Companion publications: The 50-State Authorization Price Atlas · The Model Data Center Authorization Code

Value propositions: MindCast 2027 National Legislative Session Data Center Value Proposition · MindCast Washington State 2027 Legislative Session Data Center Value Proposition

War game companions: MindCast 2027 National Legislative Session Data Center War Game · MindCast Washington State 2027 Legislative Session Data Center War Game

Interactive war games: National Interactive War Game · Washington State Interactive War Game

Thesis. The 2026 midterms split data center authorization in two. Local voters now decide where sites can proceed, while governors and utility commissions set the terms of power and service. The 2027 sessions will answer opposition with written conditions far more often than with pauses.

Why now. Texas’s grid audit is due December 10, legislatures convene in January and Pennsylvania plans its final curtailment order for January 28, 2027. State lawmakers · regulators · developers · hyperscalers · utilities will all work under terms written in the first half of 2027.

Prior forecasts in this series. Two forecasts have resolved as hits from AI Data Center Moratoriums Are Forecastable — and They End in Pricing Rules, Not Bans (August 2026). MC-ATX-P2 (70–80%) called at least three more Oregon localities to enact pauses before October 31, 2026: Woodburn (KPTV), Salem (KPTV) and Clackamas County (KOIN) enacted them. MC-ATX-P5 (55–65%) called at least one more state acting through the executive corridor by December 31, 2027, and Oregon’s governor paused state environmental permits for data centers on October 9 (KPTV). Misses: none resolved so far.

I. Executive Summary

The 2026 midterms put data center authorization on two linked tracks: local eligibility and statewide service terms. The governing mechanism is conditional authorization, in which tariffs, cost recovery and curtailment terms convert opposition into written conditions of service. The 2027 sessions test that mechanism across three groups of states: 29 with statewide frameworks, 10 that govern through utility tariffs alone and 11 with neither.

On November 3, voters decided 41 data center measures in ten states, and all 13 proposals to facilitate development failed. Democrats also flipped five Republican-held governorships by Associated Press calls, among them Texas, Ohio and Georgia (CNBC). The vote changed who holds the executive pen while Texas holds both its data center interconnection approvals and its environmental permits pending a state audit.

Thesis. The ballot determines where sites can proceed, while governors and regulators shape the terms of power and service. Three effects carry the change, and each runs on its own unit. Site eligibility moves by municipality, permit timing by covered permit class and service terms by contracted capacity.

What distinguishes the MindCast approach. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through MindCast AI Proprietary Cognitive Digital Twin Foresight Simulations (MP CDT FS). Behavioral Economics supplies the decision rules, including how voters weigh a visible cost against a diffuse benefit. Game Theory supplies the payoff structure among operators, utilities, hosts and approvers. Predictive simulations emerge from the combination.

What the simulation finds. Conditional authorization holds as the dominant equilibrium in every group of states. The forecast set contains 28 MindCast Simulation Predictions, and six carry the headline:

What the primer does. Part II traces the three effects through the 2026 evidence and sorts the states into three groups. Part III explains how the game evolves through dated forks, and Part IV states all 28 forecasts. Part V groups states by shared approval challenge and sets out what each decision-maker can do, Part VI follows power into contracts and permits and Part VII supplies a three-gate check for any project or bill.

What existing coverage misses. Coverage counted measures, and the primer separates their legal effects. The 25 binding restrictions (including three temporary moratoria) act on land use, not on the service terms where 2027 bargaining will happen. Voters and utilities converged on the same 25-megawatt (MW) line, so a project sized to avoid a large-load tariff can meet a ban at the same threshold. Pennsylvania tied its proposed curtailment rule to a federal program the Federal Energy Regulatory Commission (FERC) rejected eight days later, so the state must write its own priority order or wait for a revised federal design.

Different questions, not unfinished bills. Lawmakers will not pick up 2026 bills where they left off. Since most 2026 sessions adjourned, governors paused permits, commissions approved tariffs and voters wrote binding local rules. The 2027 question has moved from whether to regulate data centers to how to reconcile terms that three levels of government already wrote.

Cumulative impact. The midterm effects compound rather than add. A local size threshold, a state permit precondition and a utility tariff now act on the same project, and Pennsylvania already makes local approval a precondition of state permits. New governors and Georgia’s new commission majority arrive to terms their predecessors wrote, so each 2027 bill amends a stack of existing terms rather than filling a gap. The compounding raises the stakes of the first statute in each state, because it sets how the three levels fit together.

Stakeholder callouts

• 🏛 State lawmakers. The first 2027 statute in each state sets the template its neighbors copy.

• ⚖ Regulators. Commissions will write most 2027 terms, so a pause without a tariff schedule becomes a moratorium no one voted for.

• 🏗 Developers. Eligibility, permit timing and service terms now fail separately, so a site cleared at one gate can still stop at the next.

• ☁ Hyperscalers. Curtailment rank, not connection alone, decides what firm capacity is worth in PJM after the October 9 federal order.

• ⚡ Utilities. Staged-commitment tariffs sort speculative requests out before infrastructure spending starts.

Bottom line. The 2027 sessions will write conditions far more often than pauses, and the first decisive signals arrive by January 31, 2027.

II. The 2026 Election Moved Three Effects on Three Clocks

Three effects changed in 2026, and each has its own decision-maker and unit. Local ballots changed site eligibility, governors changed permit timing and commissions changed service terms. Behavioral Economics and Game Theory explain why executives and commissions moved first.

A. Three Questions Replace One Ranking

Authorization instruments act on projects in three distinct ways, and each effect has its own reversal path.

• Site eligibility. Whether and on what local conditions a parcel may host a data center. Local ballots, zoning and bans act here.

• Permit timing. Whether covered approvals issue and on what preconditions. Executive orders and agency directives act here.

• Service terms. What a project must commit to receive power, and who curtails first in an emergency. Tariffs, commission rules and grid-operator filings act here.

Ballot bans, permit pauses and tariffs share no denominator, so no megawatt total can rank them without matched project-level data.

Takeaway. Every project now faces three questions: where it can proceed, when its permits issue and what it must commit to receive power.

B. Eight Terms Tie Each Effect to an Instrument a Reader Can Find

Eight terms carry the primer’s argument, and each maps to an observable instrument.

• Large-load tariff. A utility rate schedule that sets what a customer above a size threshold must commit before service, such as study payments, minimum take and contract length.

• Interconnection approval. The grid operator’s or utility’s decision to connect a load to the transmission system.

• Curtailment. An order that a load reduce or stop drawing power during a grid emergency.

• Critical load. A customer class exempt from emergency curtailment, such as hospitals.

• Capacity charges. Payments for generation held available to meet peak demand, set in PJM by auction and passed through to customers.

• Use prohibition, size threshold and moratorium. A ban on the use, a ban above a stated scale and a time-limited suspension of applications.

• Voter-approval requirement. A rule that sends large projects to a referendum.

• Codified verification. Standing rules, adopted by protocol or statute, that test whether a requested load is real before approval.

Sourced statements are verified facts. Statements marked as MindCast analysis are inferences, and Simulation Predictions carry an identifier and a band. Load-bearing assumptions appear with the forecasts in Part IV.

Takeaway. Each term names an instrument a reader can find in a statute, tariff or order.

C. Voters Changed Site Eligibility Through Six Legal Forms

Local voters acted directly on site eligibility, and the legal effect varied by measure. Ballotpedia counted 41 measures on November 3, with 28 that would limit development and 13 that would facilitate it. Ballotpedia’s 2026 tally reached 47 decided measures across 13 states, with 46 favoring limits and three December measures still pending (State Capital Lobbyist). The tally counts passed restrictions, advisory votes and defeated pro-development proposals alike as outcomes favoring limits.

An outcome favoring limits is not an enacted restriction. Appendix C sorts each November 3 measure by legal form from unofficial results: 25 binding restrictions passed, 2 advisory restrictions passed, 1 restriction failed and all 13 pro-development measures lost.

• Use prohibitions. Alhambra (CA), Edgewater (FL) and Ohio municipalities including Granville · Piqua · Urbana removed data centers as a permitted use.

• Size thresholds. Conneaut, Defiance, Oregon and Sunbury in Ohio and Ottawa (KS) prohibited facilities above 25 MW.

• Temporary moratoria. Peru, Wells and Wiscasset in Maine suspended applications for six months or 180 days.

• Voter-approval requirements. Grove City (OH) and Janesville (WI) added a referendum gate for large projects.

• Incentive conditions. Perry (OH) required financial guarantees for hyperscale incentives.

• Project decisions. Wilmington (OH) voters rejected four rezonings tied to one project, and Ashville (OH) voters backed blocking a campus already under construction (Fortune).

Harwood (ND) voters went the other way and rejected added review (State Capital Lobbyist) while a campus rose nearby (KVRR). Ballot law also amends slowly, because a charter amendment or initiated ordinance generally requires another vote to change.

Takeaway. Binding measures act locally and by legal form, so each one needs measure-by-measure reading before a site decision.

D. Communities Moved From Participants to Holders of an Independent Authorization Gate

The November measures changed the role communities play. Data center regulation has expanded from a state–utility bargain into a three-level authorization system that includes communities as independent decision-makers over site eligibility.

Communities moved in three ways.

• From council rules to direct legislation. Voters in 25 jurisdictions enacted binding rules themselves rather than leaving them to councils.

• From influence to eligibility. Communities moved from influencing conditions to determining site eligibility.

• From hearings to classes. Communities moved from project-specific hearings to class-wide bans, thresholds and voter-approval requirements.

The local electorate now stands as a separate decision-maker alongside governors, legislatures and utility commissions. A regulator can approve service terms and a governor can issue permits, yet a community can still make the parcel ineligible. Harwood shows the new gate is not inherently restrictive, because voters can authorize development as well as stop it.

The shift applies where state law permits meaningful local authority. The next contest concerns whether 2027 legislatures codify, constrain or preempt the new role, not whether communities matter. SH10 and LH2 through LH4 forecast that contest.

Takeaway. Communities now hold an independent authorization gate over site eligibility, and 2027 legislatures decide how far state law lets it reach.

E. Governors Changed Permit Timing

Executive instruments acted on permit timing statewide, and the leading examples predate the election. Texas holds two separate gates. Governor Abbott directed the Public Utility Commission of Texas (PUCT) and the Electric Reliability Council of Texas (ERCOT) on June 10 to review the impact of data centers on electric infrastructure and residential ratepayers. His August 3 directive paused Batch Zero, ERCOT’s first group of large-load interconnection studies (Greenberg Traurig).

His September 21 letter directed the Texas Commission on Environmental Quality (TCEQ) to pause the issuance of all permits related to data center projects until ERCOT completes its review (Beveridge & Diamond). The letter sets no megawatt threshold and no firm endpoint (National Law Review). ERCOT officials aim to file the audit report by December 10, and that report informs a resumption decision rather than authorizing one. ERCOT’s large-load queue exceeds 470 gigawatts (Inside Climate News), so requested load cannot be read as committed load.

Pennsylvania’s Executive Order 2026-05 directs its Department of Environmental Protection (DEP) to review permits for projects above 25 MW only after local approval (Morgan Lewis). The order cited more than 100 proposed projects, of which 20 had sought DEP permits. The order removes data center projects from the Permit Fast Track program, while projects that sign a GRID consent order gain rolling review and Permit Decision Guarantee eligibility. The order adds a precondition and a defined path at once.

New York paused discretionary state environmental permits for data centers of 50 MW or more on July 14 (Davis Wright Tremaine). Oregon followed on October 9 (KPTV).

The vote then changed who holds the executive pen in 2027. Hinojosa and Bottoms each made data center regulation a key part of their campaigns (Texas Public Radio; AJC). Ohio’s governor-elect proposed during the campaign a conditional moratorium requiring data centers to pay their own power costs without raising other customers’ rates and to use union labor (Signal Ohio).

Takeaway. Governors set permit timing in 2026, and the 2027 governors in the largest affected states campaigned on writing conditions.

F. Commissions Changed Service Terms

Service terms decide what a project must commit to receive power. AEP Ohio’s data center tariff moved its request pool through successive stages. Of the initial 30,000 MW of requests, 13,022.7 MW paid for an engineering study and 5,642 MW signed binding contracts under the tariff (AEP Ohio). The stages show screening and commitment; they do not show why each request that failed to advance dropped out.

States wrote cost assignment into law and rule between August and October 2026.

• California. Seven laws signed September 21 include a tariff regime under the state utility commission, with tariffs due by January 1, 2028 (National Law Review).

• Alabama. Act 2026-610 requires utilities to show the Public Service Commission that large-load data center contracts recover incremental costs (ABC 33/40).

• New Jersey. S3379/A4096 adds semiannual energy and water reporting (New Jersey Governor).

• Nationally. The Edison Electric Institute (EEI) counted 25 states with an approved large-load tariff and 8 with tariffs pending (EEI).

Georgia voters gave Democrats a 3-2 majority on the Public Service Commission (Georgia Public Broadcasting), after races that turned on power bills, with criticism centered on Georgia Power’s data center contracts (AJC). FERC on October 9 accepted PJM’s Large Load definition and its Large Load Registry but rejected the Interim Resource Adequacy Service package (FERC order). FERC found that several provisions appeared to reach retail customers and retail rates. Important retail-load questions now proceed in state forums, while federal transmission and wholesale rules continue to shape the outcome.

Takeaway. The instruments written since August condition power on commitment and also define how a committed project gets served.

G. Twenty-Nine States Have Written Terms, Ten Rely on Tariffs and Eleven Have Neither

A state counts as a framework state when a statute, statewide rule or executive order binds large-load service, siting or local authority. Tax incentives and studies do not count. The 50-State Authorization Price Atlas (2026) supplies the coding behind the groups.

The Atlas coded 22 framework states, and seven more enacted statutes binding large-load service, siting or local authority in 2025 and 2026. Six appear in the Start Business by State tracker: Delaware · Idaho · Indiana · Nebraska · Tennessee · West Virginia. Maryland’s 2026 law lowered its large-load threshold to 25 MW (Maryland Office of People’s Counsel).

• Framework states (29). AL · AZ · CA · DE · FL · ID · IL · IN · MA · MD · ME · MN · MO · NC · NE · NJ · NY · OH · OK · OR · PA · SD · TN · TX · UT · VA · WA · WV · WY.

• Tariff-only states (10). GA · IA · KS · KY · LA · MI · NV · ND · SC · WI.

• States with neither (11). AK · AR · CO · CT · HI · MS · MT · NH · NM · RI · VT. Montana and New Mexico have first tariffs pending.

Takeaway. Framework states choose how far to tighten, tariff-only states choose whether to legislate and states with neither choose whether to act at all.

H. Loss Aversion and First-Mover Advantage Explain Who Moved First

Behavioral Economics explains why regional power costs became a local voting issue. PJM’s market monitor attributes $29.4 billion of capacity charges across four auctions to data center load, or 46.2% of the total (Monitoring Analytics). Voters weigh a visible cost increase more heavily than a diffuse benefit, and approvers answer with conditions rather than bans.

Reference points matter too: Ashville voters judged a campus already under construction by their experience of it, not by its sunk cost. Governors declare support for investment while writing preconditions that sort projects by commitment, and the gap tells developers which commitments buy a permit.

Game Theory explains the order of moves. Executives act on a shorter clock and can condition rather than prohibit, while legislatures enter 2027 bargaining over terms others already wrote. Statewide authorization is therefore the dominant fork, and Texas supplies the earliest test.

Under resumption on the audit’s technical basis, speculative queue entrants leave first because verification demands deposits and load data they cannot supply. Developers with financed load then re-enter under curtailment protocols, since Texas’s 2025 large-load law (SB 6) already requires transmission-voltage loads interconnecting after 2025 to accept them (Compute Law). Resumption on a ratepayer cost-recovery basis would bring ratepayer advocates and municipalities into commission dockets and make the cost terms portable to Ohio, where the governor-elect has proposed that frame.

Takeaway. Loss aversion concentrates opposition on visible costs, and first-mover advantage puts executives and commissions ahead of legislatures.

Conclusion. Ballots, governors and commissions each moved a different effect in 2026, and the 2027 sessions take up all three.

• 🏛 State lawmakers. States counted as rule-free by tariff surveys may already carry statutory terms that bind the next bill.

• 🏗 Developers. A size threshold, a moratorium and a voter-approval rule each close a site differently, so legal form decides the next step.

III. The Game Evolves Through Dated Forks Toward Conditions

The election installed two linked games, and each resolves through forks with known dates. Rules written before the sessions hold through the forecast window, so the forks decide form rather than direction.

The forecast horizon runs from November 2026 through July 2028. The forums are state executives, state utility commissions, grid operators and municipal electorates. FERC sets the federal boundary, and the primer excludes federal siting legislation, the outcome of any single pending permit and the 2028 elections.

A. Rules Written Before the Sessions Anchor the Forecast Window

Three moves fixed most of the rules before the 2027 sessions open. Executives and commissions wrote first, FERC returned retail curtailment questions to the states and the November 3 ballot set local eligibility in 25 jurisdictions. Four forces keep those rules in force through the forecast window.

• Statutory inertia. Twenty-nine states have terms that a session must amend to change.

• Amendment friction. Voter-enacted restrictions change only by another vote.

• Federal restraint. FERC most likely leaves state choices open through January (80%).

• Session calendars. Legislatures move only inside their sessions.

Validity conditions. The forecasts stay valid while two exceptions remain dormant: an unscheduled reliability or rate shock that multiplies pauses, and an Ohio statute that displaces at least five of the 22 voter-enacted restrictions. The second exception is called here the Ohio displacement statute. If either fires, the affected forecasts no longer hold as stated.

Takeaway. The forecasts rest on rules written before the sessions, and two modeled events carry the largest risk of replacing the equilibrium.

B. Five Competing Hypotheses Frame the Contest

Eleven Cognitive Digital Twins (CDTs) model the decision-makers: Texas executives and agencies · Pennsylvania’s executive and commission · the Ohio and Georgia governors-elect · state commissions · FERC and PJM · utilities · developers and hyperscalers · local electorates · state legislatures. Five hypotheses and a no-change alternative frame the simulation, each with its own falsifier.

• H1 · Conditional authorization dominates. Falsifier: new statewide pauses or bans reach half the count of new conditional instruments in the first half of 2027.

• H2 · Texas re-enters by agency protocol first. Falsifier: a statute or new executive order resumes approvals first, or none resume by July 31, 2027.

• H3 · States set curtailment rank before FERC. Falsifier: FERC accepts a revised PJM design supplying the priority before any state acts.

• H4 · Local authority reaches statehouses. Falsifier: no bill to codify or preempt local siting authority is introduced by July 31, 2027.

• H5 · New majorities condition rather than pause. Falsifier: Georgia’s commission pauses interconnection before it conditions any contract.

• H0 · No change. Falsifier: any of H1 through H5 resolves as stated.

Takeaway. Each hypothesis can fail on a named event, so the forecasts in Part IV test them rather than assume them.

C. Five Current Forks Resolve Between January and July 2027

The fork tree orders every fork by the date it resolves. Current forks carry Snapshot odds in tens, and later forks link to the forecasts in Part IV.

Three mechanisms decide how the forks resolve. Permit holds clear when an agency can verify which loads are real. Staged tariffs ask for money before infrastructure, so requests that cannot commit tend to leave. Council moratoria end in written terms when a path exists, while voter-enacted bans lack that path.

Takeaway. Verification, commitment and an available path to terms decide whether a hold, a request or a moratorium converts into written conditions.

Conclusion. The forks resolve on known dates, and each resolution moves named forecasts in Part IV.

• ☁ Hyperscalers. A FERC narrowing order would move curtailment from state dockets to PJM compliance filings across a whole portfolio.

• ⚡ Utilities. Each approved tariff becomes the next commission’s template, so early filers shape the terms their peers adopt.

IV. MindCast Simulation Predictions on the 2027 Sessions

The forecast set contains 28 MindCast Simulation Predictions, all original to this publication and dated November 5, 2026. Four layers carry them, and each has its own prefix.

• Snapshot Predictions (SN). Odds in tens at the current forks.

• Short-horizon Simulation Predictions (SH). The 2027 sessions.

• Conditional Simulation Predictions (C). Outcomes once a named branch occurs.

• Long-horizon Simulation Predictions (LH). Outcomes through July 30, 2028.

Every entry is an event probability with a window and a falsifier. Entries without a named verification source resolve against the sources in Part VIII.

Load-bearing assumptions. The measure set certifies as reported. Every group-based forecast counts states as grouped on November 5, 2026. Instruments count on their adoption date, and instruments in force before November 6, 2026 never count.

A. Snapshot Predictions: The State Forks Lead With Conditions

Snapshot Predictions give odds in tens that sum to 100% at each current fork.

• SN1 · States answer the election mainly through conditional authorization through July 2027 (60%). Other branches: restrictive delay and fragmentation 30% · federal recentralization 10%. Shift trigger: a reliability, rate or water shock that produces new pauses. Verification source: statutes, executive orders and commission orders.

• SN2 · Texas first resumes data center approvals through an agency protocol or rule by July 31, 2027 (50%).Other branches: executive framework or statute 20% · no class of approvals resumes 30%. Resumption means at least one class of approvals restarts under operative terms. Verification source: PUCT, ERCOT and TCEQ actions, executive orders and session law.

• SN3 · Pennsylvania adopts a state-anchored large-load curtailment priority in a final order by March 31, 2027 (60%). Other branches: priority keyed to a revised federal design 10% · deferred, omitted or delayed 30%. Verification source: Docket M-2026-3064961.

• SN4 · Georgia’s Public Service Commission conditions new or existing large-load contracts in its first large-load order by June 30, 2027 (50%). Other branches: no large-load order 40% · interconnection pause 10%. Verification source: Georgia Public Service Commission dockets.

• SN5 · Ohio’s governor-elect conditions state incentives or permits as a first data center action by April 30, 2027 (50%). Other branches: legislative proposal only 30% · executive pause 20%. Verification source: Ohio executive orders and Tax Credit Authority actions.

• SN6 · FERC issues no order narrowing state large-load choices by January 31, 2027 (80%). Other branch: narrowing order 20%. Verification source: FERC eLibrary.

Takeaway. The state forks lead with conditions rather than pauses, while FERC most likely leaves state choices open.

B. Short-Horizon Simulation Predictions: The 2027 Sessions Write Conditions

Short-horizon bands hold across the current branches unless a conditional forecast says otherwise.

• SH1 · States adopting new conditional instruments number at least twice the states adopting new statewide pauses or bans by July 31, 2027 (88–95%). C counts states adopting a new conditional instrument (tariff · cost recovery · disclosure · verification · local role) from November 6, 2026 through July 31, 2027. P counts states adopting a new statewide pause or ban of any duration. SH1 holds when C ≥ max(1, 2 × P). Falsifier: C < max(1, 2 × P). Verification source: statutes, executive orders and commission orders adopted in the window.

• SH2 · Conditions make up at least 75% of statewide data center instruments adopted in 2027 (80–90%).Falsifier: pauses or bans exceed 25% of instruments adopted by December 31, 2027.

• SH3 · Pennsylvania adopts a state-administered large-load curtailment priority before FERC accepts a revised PJM design, by July 31, 2027 (60–72%). Falsifier: FERC accepts a revised design supplying the same priority first, or Pennsylvania adopts no priority. Verification source: Docket M-2026-3064961 and FERC dockets.

• SH4 · Georgia’s commission opens a large-load cost or contract proceeding before any general interconnection pause, by July 31, 2027 (65–78%). A qualifying order also counts. Falsifier: a pause comes first, or neither action occurs.

• SH5 · At least 2 of the 10 tariff-only states add binding terms beyond their tariffs by December 31, 2027 (58–72%). Binding terms mean a statute, an executive instrument or a statewide siting or local-authority rule. Falsifier: one state or none.

• SH6 · At least 2 of the 11 states with neither a framework nor a tariff adopt a first statewide instrument by December 31, 2027 (62–76%). Instruments of either polarity count, including a first commission-approved large-load tariff. Falsifier: one state or none.

• SH7 · At least 2 of Delaware · Maryland · Montana · New Mexico approve a first large-load tariff by December 31, 2027 (78–90%). Falsifier: one approval or none. Verification source: commission orders.

• SH8 · No framework state repeals or suspends its statewide framework without replacement by December 31, 2027 (85–93%). Falsifier: any such repeal or suspension.

• SH9 · At least 5 framework states put a new binding obligation into operation during 2027 (65–78%).Qualifying obligations are curtailment rank, a water condition or a disclosure mandate. An earlier statute counts only when its new obligation becomes operative in 2027. Falsifier: four states or fewer.

• SH10 · A bill to preempt or codify local siting authority is introduced by July 31, 2027 in a state with a November binding restriction (78–88%). Qualifying states: California · Florida · Kansas · Maine · Ohio · Wisconsin. Falsifier: no qualifying bill is introduced. Verification source: state legislature bill databases.

• SH11 · At least 2 state commissions outside Texas pause or condition large-load interconnection pending tariff or rule development by December 31, 2027 (60–72%). Falsifier: fewer than two qualifying orders. Verification source: commission orders.

Takeaway. The 2027 sessions convert opposition into conditions, and the open questions concern which instrument carries the terms.

C. Conditional Simulation Predictions: Five Forks Change the Answer

Each conditional forecast applies only if its named branch occurs, and its falsifier is the outcome failing inside that branch by the stated date.

• C1 · If Texas re-enters through an agency protocol, resumed approvals require PUCT large-load registration by July 31, 2027 (80–90%). Condition: the agency-protocol branch of SN2.

• C2 · If Pennsylvania adopts a state-anchored priority, a second PJM state commission adopts a large-load curtailment priority by December 31, 2027 (45–60%). Condition: the state-anchored branch of SN3. Given the federal-keyed or deferred branch instead: 20–35%.

• C3 · If Georgia’s commission conditions contracts, it also reopens existing Georgia Power data center contracts by December 31, 2027 (25–40%). Condition: the conditions branch of SN4.

• C4 · If Ohio’s governor-elect conditions incentives or permits, Ohio enacts a statewide large-load tariff statute by December 31, 2027 (40–55%). Condition: the executive-conditions branch of SN5.

• C5 · If an Ohio statute taking effect by July 30, 2028 preempts, displaces or renders inoperative at least five of the 22 restrictions, at least 18 of the 22 still remain materially operative on July 30, 2028 (15–30%).Mechanism: once such a statute takes effect, the measures survive only if a court blocks it in time, and Ohio’s home-rule test makes a bare preemption statute vulnerable. From January 2027, a preemption statute also faces a veto from the incoming governor, and the projected House majority falls short of the three-fifths needed to override; a statute passed in the 2026 lame-duck session and signed by the outgoing governor avoids that veto (MultiState; Signal Ohio).

Takeaway. Five forks change the answer enough to need their own forecasts, and C5 carries the largest swing.

D. Long-Horizon Simulation Predictions: Restrictions Hold Unless Ohio Acts

Long-horizon forecasts cover durability, local authority, project location and ballot repetition through July 30, 2028.

• LH1 · At least 18 of the 22 non-temporary binding restrictions passed November 3 remain materially operative on July 30, 2028 (74–86%). The forecast assumes certification of the measure set. Falsifier: five or more are repealed, invalidated, preempted or bypassed. Fifteen of the 22 sit in Ohio, so the Ohio displacement statute carries most of the risk (C5).

• LH2 · At least 3 states enact statutes that expressly define, preserve, limit or preempt local data center siting authority by July 30, 2028 (78–88%). Falsifier: two states or fewer. Base rate: Florida · Louisiana · South Dakota · Virginia enacted such statutes in 2025 and 2026 (Start Business by State tracker; MultiState).

• LH3 · At least 1 state newly enacts a statute preempting local data center bans or moratoria by July 30, 2028 (30–45%). Falsifier: no state enacts one.

• LH4 · At least 1 state requires local voter approval or local consent for large data centers statewide by July 30, 2028 (20–35%). Falsifier: no state enacts one.

• LH5 · At least 6 of the next 10 announced US data center campuses of 250 MW or more sit in framework states (70–84%). Counting starts after November 5, 2026. Falsifier: five or fewer. Verification source: company announcements.

• LH6 · At least 5 new local data center ballot measures qualify across at least 3 states by June 30, 2028 (75–88%). Measures of either polarity count. Falsifier: fewer than five measures or fewer than three states. Verification source: certified ballot notices.

Takeaway. Local authority moves to statehouses mostly to be defined or preserved, and voter-enacted restrictions hold unless Ohio acts.

Conclusion. Direction remains stable across the current branches, and the open questions concern instrument, forum and the location of local authority.

• ⚖ Regulators. The mid-range bands mark where commission choices still decide the outcome.

• 🏗 Developers. Ohio preemption and a grid shock would change site strategy rather than timing.

V. What Each State Group and Decision-Maker Faces

The same forks land on states and decision-makers with different exposure. States first sort into seven clusters by the approval challenge they share. A second table states what changes for each decision-maker and what each can do now, and risk entries size the exposure in units each owner controls.

A. Seven State Clusters Share the Same Approval Challenges

States that face the same blocking gate face the same first decision in 2027, whatever their legal group. A state appears in every cluster whose challenge it faces. Ohio sits in three clusters, and seven states sit in two: Georgia · Kansas · Kentucky · Michigan · Pennsylvania · Texas · Wisconsin. Clusters follow the evidence in Part II and the groups as coded on November 5, 2026.

In MindCast analysis, clusters 1 through 4 hold the terms with dated forks pending, while clusters 5 and 6 decide whether new states write terms at all. Cluster 7 has written terms in hand, and LH5 tests whether the next large campuses favor framework states.

Takeaway. The blocking gate, not the legal group alone, sets the first decision each state faces in 2027.

B. Five Decision-Makers Face Different Consequences and Moves

Each row pairs a consequence with a move that costs little whichever branch occurs and a move that waits for a named trigger.

Texas has split legislative control, while Ohio and Iowa keep Republican legislatures (MultiState). Divided government favors agency orders, utility dockets and bipartisan provisions over sweeping statutes.

Takeaway. Every move now separates the three effects, and every trigger move names the event that releases it.

C. Five Risk Entries Size Exposure Before the Forecasts Resolve

Each entry names its forecast source, the exposure in its owner’s unit, unilateral mitigation and the exposure that remains.

Entry 1 · 🏗 Developers · Texas re-enters by agency protocol.

• Exposure. A developer that arrives unprepared waits behind verified projects. The loss runs in quarters of energization schedule per Texas project. Severity: high for Texas-concentrated pipelines. Probability: 50% on the current Texas fork (Snapshot SN2). Given that branch, resumed approvals require PUCT registration at 80–90% (conditional C1).

• Mitigating actions. The interconnection lead assembles load, financing and curtailment-acceptance data for every Texas project before the ERCOT audit report due December 10, 2026. The regulatory affairs lead files comments in any PUCT verification rulemaking before the Legislature convenes January 12, 2027. The finance lead reserves deposit capacity for the verification terms before approvals resume.

• Residual exposure. A statutory branch can add requirements no developer can meet before bill text exists.

Entry 2 · ⚡ Utilities · Tariff-only states and states with neither adopt binding terms.

• Exposure. A utility without an approved large-load tariff draws uncommitted requests as neighboring states adopt them. The utility then carries study and upgrade costs it cannot assign, measured in unrecovered upgrade dollars and rate-case quarters. Severity: moderate to high. Probability: 58–72% that at least 2 tariff-only states add binding terms and 78–90% that at least 2 pending first tariffs win approval in 2027 (short-horizon SH5 and SH7).

• Mitigating actions. The rates lead files a staged-commitment tariff modeled on AEP Ohio’s stages in the first half of 2027. The large-load account lead requires study payments before engineering work, starting with the next request. The regulatory lead documents cost causation for each existing large-load contract before the next rate case.

• Residual exposure. A new commission majority can rewrite approved terms within one cycle, as Georgia’s result shows.

Entry 3 · 🏗 Developers · Council moratoria end in written terms.

• Exposure. Sites in council-moratorium jurisdictions regain eligibility under conditions. A developer who let its options lapse there loses the site and the first position under the new terms. Severity: moderate. Probability: outside the forecast set. Moratorium expirations listed in Part VIII resolve the branch case by case.

• Mitigating actions. The land lead classifies each site by legal form before 2027 site decisions: council moratorium, voter-enacted ban or voter-approval rule. The land lead also holds options on council-moratorium sites through each expiration date. The community relations lead drafts host-agreement terms before each moratorium expires.

• Residual exposure. Written terms can carry host fees and guarantees, as in Perry, that cut site returns after eligibility returns.

Entry 4 · ☁ Hyperscalers · Commissions pause or condition interconnection.

• Exposure. Campuses in states that pause lose quarters of energization schedule. Terms written after commitment also raise the cost of firm service per megawatt. Severity: high for PJM portfolios. Probability: 60–72% that at least 2 commissions outside Texas pause or condition interconnection in 2027 (short-horizon SH11).

• Mitigating actions. The energy procurement lead files comments on Pennsylvania’s tentative order before its comment deadline. The site selection lead sequences 2027 commitments toward states with approved tariffs. The engineering lead designs a curtailable load segment into the next interconnection application.

• Residual exposure. Curtailment terms can still differ state by state across one portfolio.

Entry 5 · ⚖ Regulators · Commissions pause or condition interconnection.

• Exposure. A commission that pauses without a tariff schedule holds an open-ended docket. The cost runs in months of docket time and in interconnection requests held without a decision date. Severity: moderate. Probability: 60–72% for the same branch (short-horizon SH11). A Georgia pause rather than conditions sits at 10% (Snapshot SN4).

• Mitigating actions. Commission counsel pairs any interim order with a dated tariff schedule. Technical staff adopts a large-load reporting list, as Pennsylvania’s tentative order proposes, before the first interim order. Staff aligns state curtailment rules with its grid operator’s compliance filings after the November show-cause responses.

• Residual exposure. A governor or legislature can impose a pause the commission did not design.

Nothing in Part V constitutes legal, investment or fiduciary advice.

Takeaway. Each entry sizes exposure in the unit its owner controls, and none claims the exposure disappears.

Conclusion. Each decision-maker can act now without waiting for a fork to resolve, and every residual exposure stays named.

VI. Power Carries Local Opposition Into Contracts, Permits and Water Terms

Power links regional capacity costs to local data center permits. Five concerns carry direct evidence, Washington State shows terms arriving before a statute and five competing explanations test the thesis.

A. Five Power Concerns Move Conditions Off the Zoning Map

Each concern moves a condition from the zoning map into the utility contract, the permit or the water supply behind the campus.

• Opposition has reached storage. El Dorado (KS) voters approved a prohibition on high-impact data centers and certain battery storage with 71%.

• Texas holds onsite-generation permits with the rest. Bracewell reads the September 21 letter as covering permits without an onsite-generation exception (Bracewell).

• Pennsylvania proposes curtailment priority by rule. The October 1 tentative order proposes that utilities shed large computational loads first and that data centers not count as critical load (tentative order). A final order is planned for January 28, 2027.

• The 25 MW line appears in both tariffs and bans. Five binding November restrictions prohibit facilities above 25 MW, and the same number opens the large-load class in AEP Ohio’s tariff, Georgia Power’s tariff and Virginia’s GS-5 rate class.

• Water is becoming a permit condition. California’s AB 2469 requires a water-supply assessment for covered local approvals, and Texas made its water-use survey a condition of TCEQ water permits.

Takeaway. Power converts local opposition into contract, permit and water terms.

B. Washington State Already Writes Terms, and the Senate Decides Whether They Join

Washington State kept its governing party, and Democrats retain legislative majorities, pending certification. Terms already arrive without a statute. Seattle City Light adopted a data center class for new service of 10 MVA or more on July 21, 2026 (KUOW). Grant PUD’s commission votes December 15, 2026 on two data center rate classes, and the Utilities and Transportation Commission is drafting large-load principles in Docket UE-260162 (UTC). Eleven local moratoria run toward expiry in 2027, led by Seattle and Spokane.

The open question is whether the Legislature joins those pieces into one statute. HB 2515 passed the House 51-41 and died in Senate Ways & Means. A dedicated Washington State simulation run of November 6, 2026 gives the Senate even odds: a two-regime statute 30%, an investor-owned-only version 20% and a hold 50%. The same run places no statewide pause by July 31, 2027 at 85–92% and a UTC final policy statement or rulemaking by March 31, 2027 at 70–82%. The Department of Revenue still holds the Data Center Workgroup’s final report pending tribal consultation (Washington State DOR), and the run gives the report 30% to arrive before the session. The full branch odds appear in the Washington State war game. The Washington State figures sit outside the 28 national forecasts in Part IV.

Texas, Ohio and Georgia now carry the unknown-terms profile the Atlas assigned to Washington State.

Takeaway. Washington State and utilities such as Puget Sound Energy can file against the cost-causation principle now, because a Senate hold would leave boards, the commission and councils writing terms one utility at a time.

C. Five Competing Explanations Each Name What Would Weaken the Thesis

Five alternatives test the thesis, and each names an observable that would weaken it.

• The election turned on other issues. The primer claims salience rather than causation, and the executive instruments that set timing predate the vote.

• Local opposition stays local. Codification or state preemption would each move eligibility decisions to the state level and weaken the thesis.

• Tariffs deter rather than sort. AEP Ohio’s stages fit both readings, because the data do not show why requests dropped out. If tariff states lose committed projects to tariff-free states, the sorting mechanism fails even where tariffs spread.

• Federal recentralization. A wholesale large-load rule from the FERC show-cause dockets could narrow state choices without preempting any statute.

• Supportive electorates. Harwood shows voters accepting a campus they can see operating. If more towns follow, eligibility loosens without state action.

Takeaway. Each alternative names an observable that would weaken the thesis, and none currently displaces the central thesis.

Conclusion. Executives and commissions now write the terms, and power carries them into contracts, permits and water supply.

• ⚡ Utilities. The 25 MW collision means tariff thresholds now interact with local bans at the same line.

VII. Any Project or Bill Can Be Tested Against Three Gates

The three effects convert into a check that a lawmaker, regulator, developer or utility can run on any project or bill. Each gate names a question, an observable and a consequence.

A. Eligibility, Timing and Service Terms Form Three Separate Gates

• Gate 1 · Eligibility. Which legal form governs the parcel? Observable: the measure text and its effective date. Consequence: a size threshold leaves smaller designs eligible, and a moratorium returns eligibility at expiration unless extended.

• Gate 2 · Timing. Which authority holds the permit, what pauses it and what lifts the pause? Observable: the directive’s text and its stated endpoint. Consequence: Texas’s interconnection and permit pauses rest on different instruments, so a project needs both lifted.

• Gate 3 · Service terms. What must the project commit before it receives power, and who curtails first? Observable: the approved tariff and the state’s load-control rules. Consequence: commitment requirements and curtailment rank set the value of firm service.

The three-gate check assumes the gates operate separately, so clearing one gains nothing at the others. Pennsylvania’s order already links the first two by making local approval a precondition of state permits, and a statute bundling all three would change the check.

Takeaway. Each gate has its own decision-maker, and a project clears authorization only when all three clear.

B. Three Rules Carry the Election’s Lessons to Any Contest

Three rules generalize the election’s lessons for lawmakers, regulators and project teams.

• Forum authority. Forecast an approval at the forum that holds authority over the cost being allocated. FERC found PJM’s curtailment design reached retail jurisdiction, so states now write the priority.

• Threshold collision. Where a ban threshold and a tariff threshold share a number, a design sized below one line meets the other. Thomas Schelling’s The Strategy of Conflict (1960) explains the convergence: parties coordinate on prominent numbers without negotiating. The transfer holds because voters and commissions both need a bright line a lay reader can apply, and it fails where thresholds come from engineering studies.

• Amendment friction. Instruments enacted by ballot change only by another vote, so they sit outside negotiated conversion paths such as council moratoria.

Takeaway. The rules tell lawmakers and regulators where to set thresholds and which forum to watch.

Conclusion. The gates and rules let any reader apply the primer to a project or bill it does not name.

• 🏛 State lawmakers. A bill that bundles eligibility, timing and service terms changes all three gates at once and should say so.

VIII. What to Watch: Texas Re-Entry Is the First Decisive Fork

Statewide authorization is the dominant fork, and Texas re-entry resolves first after the ERCOT audit report due December 10. Each item names its date, source and the forecasts it moves.

• FERC show-cause responses · November 16 and 20, 2026. Source: FERC eLibrary. Moves SN6 and SN1.

• ERCOT audit report · by December 10, 2026. Source: ERCOT. Moves SN2 and C1.

• Governors-elect first actions · from January 2027. Source: executive orders. Moves SN2, SN5 and SH1.

• Pennsylvania final order · planned January 28, 2027. Source: Docket M-2026-3064961. Moves SN3, SH3 and C2.

• Bill filings in all three groups · weekly during sessions. Source: state legislature databases. Moves SH1 · SH2 · SH5 · SH6 · SH8 · SH9 · SH10 · LH2 through LH4.

• Georgia commission dockets · monthly. Source: Georgia Public Service Commission. Moves SN4, SH4 and C3.

• Ohio tariff and preemption bills · weekly in session. Source: Ohio General Assembly. Moves C4, C5, LH1 and LH3.

• Large-load tariff approvals · monthly. Source: EEI survey and commission orders. Moves SH7 and SH11.

• Ballot certification · through December 2026. Source: county and municipal election authorities. Moves LH1 and corrects Appendix C.

• California laws taking effect · January 1, 2027. Source: California statutes and commission decisions. Moves SH9.

• Local moratorium expirations · at each expiration date. Source: local ordinance texts. Resolves risk Entry 3 case by case.

• New York Executive Order 62 expiration · by July 14, 2027. Source: New York executive orders. Moves SH1.

• New local ballot measures and campus announcements · quarterly. Source: election authorities and company announcements. Moves LH5 and LH6.

Takeaway. The ERCOT audit report, the Pennsylvania final order and January’s executive actions supply the first decisive updates.

Conclusion

Local voters prohibited, paused or conditioned data center development where binding measures passed. Governors paused and conditioned state permits, and commissions moved request pools into contracted commitments. Each effect operates on its own unit, and none reduces to a single ranking. Communities now hold an independent authorization gate alongside governors, legislatures and commissions.

The simulation finds that the 2027 sessions will channel opposition through power cost, reliability, disclosure and local-approval conditions. States adopting new conditions will number at least twice the states adopting new pauses by July 31, 2027 (88–95%). Two modeled events carry the largest risk of replacing the equilibrium: the Ohio displacement statute and a grid shock that multiplies pauses.

If states adopting new conditions fall below twice the states adopting new statewide pauses by July 31, 2027, the conditional-authorization thesis fails. Twenty-eight MindCast Simulation Predictions carry the claim, led by SH1 (88–95%), SN2 (50%) and LH1 (74–86%).

Working With MindCast

The forecasts in this paper describe the national board, and no national forecast can tell one developer, utility or state which terms to seek in 2027. The answer turns on facts only the client holds: where its project sits in the queue, how the project is financed, which forum decides and who sits across the table.

MindCast takes those facts in confidence. MindCast builds Cognitive Digital Twins of the institutions that decide the client’s terms: the utility and state commission, the governor’s office and the host community. The simulation then runs the 2027 contest against the options the client actually has. Each engagement is a Live-Fire Mission, meaning MindCast tests one live decision before the client commits to it.

Five engagements cover the decisions most clients face in 2027:

• Developers waiting to re-enter Texas. The mission tests how the ERCOT audit report due December 10 and the rules that follow it treat the client’s place in the queue.

• Hyperscalers and utilities with campuses in PJM states. The mission tests what Pennsylvania’s final curtailment order, planned for January 28, does to the client’s campus portfolio. The mission also tests whether neighboring PJM states copy the order.

• Utilities and commissions writing their first data center terms. The mission drafts those terms from The Model Data Center Authorization Code and tests how developers, ratepayers and lawmakers respond.

• Lawmakers and local governments deciding who controls siting. The mission tests bills that preempt, codify or preserve local authority against the measures voters passed on November 3.

• Clients with projects or a legislative agenda in specific states. MindCast builds a deck for each state on retainer. Each deck covers the state’s group, the rules already in force, who decides and which forecasts in this paper would move the state. Individual state war games are also available on retainer.

The companion pieces show the method at work. The national value proposition and the Washington State value proposition describe each engagement and what it delivers. The national war game lays out how six national events could branch across fifteen states, and the interactive war game lets readers play those turns themselves. The Washington State war game does the same for one state, and the Washington State interactive war game lets readers play it from the seat of a legislative committee, a utility or a city council.

Contact mcai@mindcast-ai.com.

Appendix A. Related MindCast Work

2027 Legislative Session Data Center Series

• MindCast 2027 Legislative Session Data Center Value Proposition (Q4 2026). Rates the three groups of states on the Model Code’s twelve levers and sets out the engagements keyed to the forecasts.

• MindCast Washington State 2027 Legislative Session Data Center Value Proposition (Q4 2026). Applies the primer to Washington State’s two utility regimes and council moratoria.

• MindCast 2027 Legislative Session Data Center War Game (Q4 2026). Plays six national events and their successor states from three seats, with state overlays drawn from the Atlas and the Model Code.

• MindCast Washington State 2027 Legislative Session Data Center War Game (Q4 2026). Plays six Washington State events and thirteen successor states from three seats.

• Interactive War Game (Q4 2026). Runs the national war game’s turn sequence online.

• Washington State Interactive War Game (Q4 2026). Runs the Washington State war game’s turn sequence online from three seats.

Frameworks and methods

• The Calculus of Predictive Game Theory Evolution and Mutation (2026). Supplies the fork-tree method used in Part III.

• The Authorization Market: Standardized Bargaining, Rationed Power, and the Competition to Build America’s AI Infrastructure (August 2026). Supplies the finding that connected power is not firm power.

Prior simulations and reference resources

• The 50-State Authorization Price Atlas (PDF, 2026). Supplies the instrument-by-instrument coding behind the three groups of states.

• AI Data Center Moratoriums Are Forecastable — and They End in Pricing Rules, Not Bans (August 2026). Supplies the moratorium-to-terms mechanism behind the long-horizon forks and the two resolved forecasts reported at the top.

• The Model Data Center Authorization Code (PDF, 2026). Supplies the form-blind coding applied to ballot outcomes in Appendix C.

Related analysis

• AI Data Center Veto Points Show Where Projects Get Stopped and How Authorization Risk Moves (September 2026). Maps the approval stages where the three effects enter.

• Burden-Based Authorization for AI Data Centers, Modular Clean Compute, and the Backlash Gap Between Grid Tariffs and State Law (September 2026). Supplies measured burden as the alternative to facility labels.

Appendix B. External Sources

Primary legal, regulatory and company sources

• FERC Order, Docket ER26-3515 (October 9, 2026). Returns retail curtailment questions to state forums.

• Pennsylvania PUC Tentative Order, Docket M-2026-3064961 (October 1, 2026). Proposes the curtailment priority behind SN3 and SH3.

• Monitoring Analytics, Market Monitor Report to the PJM Members Committee (July 27, 2026). Attributes capacity charges to data center load.

• AEP Ohio: data center load under contract. Documents the staged request pool behind the sorting mechanism.

• New Jersey Governor: S3379/A4096. Adds energy and water reporting to New Jersey’s rate-class law.

• EEI: Large Load Projects and Tariffs, October 2026. Counts approved and pending large-load tariffs by state.

• Washington State DOR: Data Center Workgroup. Reports the status of Washington State’s workgroup report.

• Ballotpedia: November 3 data center ballot measures. Lists the 41 measures sorted in Appendix C.

Law-firm and policy analysis

• Beveridge & Diamond: Texas permitting pause. Quotes the September 21 directive to TCEQ.

• Bracewell: Texas TCEQ pause. Reads the pause as reaching onsite generation.

• Greenberg Traurig: Texas data center permitting halt. Describes the June 10 review directive and the August 3 Batch Zero pause.

• National Law Review: Texas pauses all environmental permitting for data centers. Notes the letter’s missing threshold and endpoint.

• Morgan Lewis: Pennsylvania Executive Order 2026-05. Describes the Pennsylvania order’s 25 MW threshold, local-approval precondition and project count.

• National Law Review: California data center laws. Summarizes California’s seven 2026 laws.

• Davis Wright Tremaine: New York Executive Order 62. Describes New York’s July 14 permit pause.

• Maryland Office of People’s Counsel: Utility RELIEF Act updates (July 2026). Describes Maryland’s 25 MW threshold and registry.

• Compute Law: ERCOT large-load interconnection. Describes Texas’s curtailment requirement for new transmission-voltage loads.

• State Capital Lobbyist: Data Centers on the Ballot. Reports Ballotpedia’s full 2026 tally and the Harwood vote.

• MultiState: State Legislatures, 2026 results. Projects 2027 legislative control by state.

• MultiState: Local data center regulations gain ground. Describes South Dakota’s local-authority statute in the LH2 base rate.

• Start Business by State: data center law tracker. Lists the 2025 and 2026 statutes behind the group recount and the LH2 base rate.

Books

• Thomas C. Schelling, The Strategy of Conflict (Harvard University Press, 1960). Explains why parties coordinate on the 25 MW line.

Reporting

• Fortune: Ohio data center ballot measures. Reports the Wilmington and Ashville votes.

• Texas Public Radio: Hinojosa’s challenge to Abbott. Reports data centers at the center of Hinojosa’s campaign.

• AJC: Where Jackson and Bottoms stand on data centers. Reports Bottoms’s data center positions.

• KVRR: Harwood data center construction. Reports construction underway near Harwood.

• MultiState: Ohio 2026 election results. Projects the Ohio House majority behind C5’s veto step.

• Signal Ohio: What comes next after the 2026 midterms. Reports the loss of the veto-proof majority.

• CNBC: 2026 midterm election results. Reports the five governorship flips.

• Signal Ohio: Acton data center plan. Describes the Ohio governor-elect’s proposal.

• Inside Climate News: ERCOT audit. Reports the size of the ERCOT queue.

• ABC 33/40: Alabama PSC review rules. Reports Alabama’s large-load contract rules.

• Georgia Public Broadcasting: PSC majority. Reports the Georgia commission’s new majority.

• AJC: Georgia Democrats take PSC majority. Reports the cost-allocation focus of the Georgia races.

• KPTV: Oregon pauses data center permits until 2027. Reports Oregon’s October 9 executive pause.

• KPTV: Woodburn pause · KPTV: Salem pause · KOIN: Clackamas County pause. Resolve the Oregon local-pause forecast reported at the top.

Appendix C. November 3 Ballot Measures by Legal Form and Operative Effect

Unofficial results as reported through November 5, 2026, pending certification by each election authority. Source: Ballotpedia via State Capital Lobbyist. Polarity follows Ballotpedia’s yes-vote classification, and each operative effect requires a check against the measure text and its effective date.

Yes would limit development (28 measures)

Binding, passed (25):

• Use prohibitions (7). Alhambra (CA) ban · Edgewater (FL) charter ban · Granville · Hubbard · Piqua · Upper Sandusky · Urbana (OH) bans. Operative effect: data centers lose permitted-use status in the jurisdiction.

• High-impact or size prohibitions (10). El Dorado (KS) high-impact ban, which also reaches certain battery storage · Emporia (KS) high-impact ban · Ottawa (KS) ban above 25 MW · North Berwick (ME) size ban · Conneaut · Defiance · Oregon · Sunbury (OH) bans above 25 MW · Pataskala and Trenton (OH) large-scale bans. Operative effect: facilities above the stated scale lose eligibility; smaller facilities remain eligible.

• Temporary moratoria (3). Peru, Wells and Wiscasset (ME). Operative effect: applications are suspended for the stated period, and eligibility returns at expiration unless extended.

• Voter-approval requirements (2). Grove City (OH) · Janesville (WI). Operative effect: large projects require a referendum; no site closes.

• Incentive conditions (1). Perry (OH). Operative effect: hyperscale incentives require financial guarantees; no site closes.

• Township zoning (2). Monroe and Sprigg (OH). Operative effect: township zoning governs data center siting, with the effect set by each zoning text.

Advisory, passed (2): Bloomington (IL) · Morgan County (IL). Operative effect: none on any permit.

Failed (1): Harwood (ND) added review. Operative effect: no change.

Yes would facilitate development (13 measures, all defeated)

• Binding project or land decisions (6). Wilmington (OH) four rezonings: the project parcels keep their prior zoning · Ashville (OH) term summary: voters backed blocking a campus under construction, with the effect on the campus set by its underlying agreement · Boulder City (NV) land opening: the city land stays closed to the project.

• Advisory (7). Chouteau · Granite City · Joliet · Nameoki · Venice (IL) · Downe Township (NJ) · Monroeville (OH). Operative effect: none on any permit.

Companion to FTC Investigation of Anthropic and OpenAI · Agentic Foreseeability and the Modern Duty of Care for AI Agents · OpenAI State Attorney General Lawsuits: Two Pleading Templates Across Ten State Courts Decide Which Game Theory Evolution Arrives First · Who Writes AI Insurance Coverage Conditions · Who Absorbs Correlated AI Loss

I. Executive Summary

One piece of evidence now serves four institutions at once. A single AI alert can show a regulator that a company detected a risk, give a plaintiff notice, expose an undertaking the company left unanswered and hand a carrier a coverage defense. The same entry cuts four ways, and the deployer controls only one variable: the response attached to it.

Common-law liability rests on three questions: who knew, who controlled and who authorized. AI agents scatter all three answers. Knowledge sits in logs no person read, control passes from developer to integrator to deployer, and authority lives in permissions held by software the law does not treat as an agent.

The thesis: four doctrines each break in a predictable direction once agents act, and the breaks collide. Voluntary undertaking, imputed knowledge and apparent authority each expand deployer liability. Res ipsa loquitur narrows a plaintiff’s proof route. The collision then runs through four stages, from naming a deployer to deciding who keeps the loss.

The four stages run on two clocks. The renewal clock moves coverage by policy date and market segment. The litigation clock moves claims by filing and ruling. The party that controls an agent, the party a plaintiff names and the party that retains the loss can differ, and the two clocks decide when each deployer learns which one it is.

Carriers sit on both sides of the board. Carriers insure deployers, sue AI vendors through subrogation and defend claims over their own AI tools. Frontier AI firms face suits as self-deployers today and as upstream vendors next, while other AI firms carry thinner balance sheets that push claims toward richer partners. Lawmakers set the rules, state attorneys general enforce them and consumer groups supply much of the evidence both sides litigate over.

MindCast applies Predictive Behavioral Economics + Dynamic Game Theory to the seats that apply these doctrines first: carriers at renewal, courts and state attorneys general in the first agent cases and deployers deciding what to monitor, retain and revoke. The paper takes the form of a Simulation Snapshot, whose many short-horizon MindCast Foresight Simulation Predictions test how agentic consumer incidents and legal incidents move. Federal enforcement in FTC Investigation of Anthropic and OpenAI enters as an accelerator rather than a gate.

The paper names the governing equilibrium and the two classical gates of foreseeability and proximate cause. The paper works through the four doctrines, six competing duties and the statutes that move them. Attribution by conduct, the four-stage chain, the two clocks and a stakeholder board come next. An incident baseline, six rival accounts and 26 Snapshot Predictions across 20 hypotheses lead to event transitions, risk mitigation and dated checkpoints.

• 💼 Carriers. The same vendor contract decides a carrier’s position as insurer and its subrogation rights as plaintiff, so underwriting should read it first.

• 💼 Frontier AI firms. Exposure arrives directly through consumer suits and attorneys general now and indirectly through deployer tenders and carrier subrogation next.

• 💼 Other AI firms. Thin balance sheets turn insolvency into a liability exit, which pushes claims toward frontier partners and insurance.

• 💼 Executives at deployers. Coverage can change at renewal before any complaint arrives, so the policy schedule is a liability document.

• 🏛️ State lawmakers. The executive order spares child-safety laws only from its proposed preemption statute and not from Department of Justice suits, while Colorado now allocates fault between developers and deployers by statute.

• 🏛️ State attorneys general. Consumer-protection law already reaches self-deploying developers, and the police-power exception can keep enforcement moving through bankruptcy.

• 🏛️ Federal lawmakers. A federal product-liability statute would expand exposure, and preemption by litigation would shrink state regimes.

• 🛡️ Consumer groups. Published test results become documented foreseeability, so advocacy evidence shapes both tort notice and coverage defenses.

• ⚖️ Counsel. The named defendant and the final loss-bearer can differ, and indemnity terms decide the gap between them.

• 📊 Investors. Retained AI loss surfaces first at portfolio companies whose renewals exclude AI cover before any claim arrives, so the renewal schedule becomes a diligence item.

II. One Alert, Four Readers: The Response Decides Exposure

The governing equilibrium is the set of answers the common law gives before it assigns liability. Agents disturb those answers.

The common law asks three questions. Who knew of the risk, who controlled the instrument and who authorized the act? Human institutions answer each one with a person, a moment and a file.

AI agents change all three answers at once. Knowledge accumulates in evaluation results and logs faster than any reviewer reads them. Control passes in sequence through a developer, an integrator and a deployer. Authority lives in permissions, and the software holding them is not a legal agent.

Duties compete when compliance with one doctrine raises exposure under another. The observable sign is a single act that reads as reasonable care under one rule and as a source of liability under a second. Monitoring, retaining logs and revoking permissions each produce that sign, which is why competing duties govern deployer strategy.

The clearest case is a single alert read by four institutions. Each reader asks a different question of the same entry, and each answer moves a different exposure.

A regulator asks whether the company detected the risk, and the alert becomes evidence of a working detection program. A tort plaintiff asks whether the company knew, and the alert becomes notice. An undertaking claimant asks whether the company answered what it chose to monitor, and the alert becomes proof of an unanswered undertaking. A carrier asks whether the insured knew before binding, and the alert becomes a known-loss or prior-knowledge defense.

The foreseeability paradox sits at the center of the competing duties. Tort doctrine turns a dated finding into notice. Coverage doctrine turns the same finding into a reason the loss was not fortuitous. A deployer who documents a risk and leaves it open increases its liability and weakens its cover in one act.

Some facts are already known. California Civil Code section 1714.46 bars the autonomy defense from January 1, 2026, and carriers have filed generative-AI exclusions built on Verisk’s standard language. The National Association of Insurance Commissioners (NAIC) names “silent artificial intelligence” alongside silent cyber in its 2026 Report on the Cybersecurity Insurance Market. At least ten state attorneys general sued chatbot developers in 2026.

Other facts are discoverable: each deployer’s logs, evaluation results, permission grants and the response attached to each open finding. What stays unknowable until a court or carrier acts is which doctrine the first reasoned ruling applies to an agent’s act, and which terms carriers write as each policy renews.

The deciding variable in every doctrine is the response attached to a finding, not the finding alone.

III. Federal Enforcement Accelerates the Common Law

Federal enforcement and the common law run on separate tracks that meet. A federal order speeds the private pathways without gating them.

FTC Investigation of Anthropic and OpenAI ranks four candidate federal theories: deception about safety, information security and incident response, unfairness from first-party agent conduct and unfairness resting on a pre-release testing duty.

The pathways run before any federal order. Raine v. OpenAI (2025) already pleads strict product liability, negligence and failure to warn in San Francisco Superior Court. Kentucky sued Character Technologies under state consumer and privacy law in January 2026, and eight more states sued OpenAI in their own courts on October 14, 2026.

Carriers re-decide AI terms as each policy renews, whatever the agency does. The FTC paper names the same channel: a coverage condition written before any order becomes the market term the order later confirms.

A federal order accelerates the pathways by supplying a borrowable basis. A deception order makes representations the live question. An unfairness order resting on agent conduct makes conduct the live question, and conduct is where the four doctrines operate. Private plaintiffs and state attorneys general borrow whichever basis arrives first, because borrowing lowers their evidentiary cost.

The common-law pathways are already moving, and the federal basis decides which one plaintiffs reach for first. Every pathway still passes two classical gates.

IV. Foreseeability and Proximate Cause Both Open Toward Liability

Every agentic claim passes two classical gates before any of the four doctrines matters. Foreseeability decides whether a duty exists, and proximate cause decides whether the breach reaches the harm. Agents change what a defendant can foresee and what counts as an intervening act. Agentic Foreseeability and the Modern Duty of Care for AI Agentssupplies the foreseeability assessment the four agentic doctrines extend.

Oliver Wendell Holmes Jr. set the external standard in The Common Law (1881). Liability turns on what would be blameworthy in “the man of ordinary intelligence and prudence,” not on what the defendant actually foresaw. Benjamin Cardozo then made foresight the boundary of duty. In MacPherson v. Buick Motor Co. (1916) a foreseeable danger carried the maker’s duty past the dealer to the buyer, and in Palsgraf v. Long Island Railroad Co. (1928) “the risk reasonably to be perceived defines the duty to be obeyed.”

MacPherson matters most for the AI stack. Cardozo held that an intermediary between maker and user does not cut the maker’s duty when the danger is foreseeable. A deployer standing between a developer and the user therefore shields the developer only as far as the danger was unforeseeable.

Holmes’s first lecture also traces early law that proceeded against the offending animal or object itself. The common law later moved liability onto the owner of the instrument. The Ninth Circuit’s description of an AI agent as “a tool, not a person for statutory purposes” in Amazon v. Perplexity (2026) keeps that modern allocation for computer-access claims, and the court expressly reserved the tort question.

Judge William Andrews, dissenting in Palsgraf, called proximate cause a line drawn by “practical politics” rather than logic. The Wagon Mound (No. 1) (1961) drew the line at harms of a foreseeable type, whatever the manner of their arrival. An agent may reach a harm by a path no engineer predicted, yet the type of harm often sits in the developer’s own evaluation results.

Defendants answer with the superseding-cause argument: the agent’s autonomous choice or the user’s circumvention broke the chain. California Civil Code section 1714.46 bars the autonomy version from January 1, 2026. Subsection (c) preserves causation and foreseeability evidence and the comparative fault of others. Defendants therefore shift to misuse and to fault among the layers of the stack, as OpenAI’s misuse argument in Raine shows.

Holmes also defined law itself as a forecast. In “The Path of the Law” (1897) he wrote that “the prophecies of what the courts will do in fact” are what he meant by the law. The Snapshot applies the same view to agentic doctrine and treats each hypothesis as a forecast of what courts, attorneys general and carriers will do. Behavioral economics adds what Holmes’s rational “bad man” omits: real actors weight the latest filing more than the base rate.

Foreseeability has no closed list. Courts judge it by the type of harm and the facts of each case, so no inventory of foreseeable harms exists for a deployer to clear. Undertaking, authority and shared control work the same way, because each turns on conduct a court reviews after the harm. A developer, a designer of an agent’s operating environment or a governance function cannot satisfy the gates by checking a list.

Agents widen what defendants can foresee and narrow what they can call an intervening act, so both gates open toward liability.

V. Voluntary Undertaking: Monitoring Creates the Duty

Voluntary undertaking is the first of the four agentic doctrines. Monitoring sits at the center of every AI safety program, and monitoring is exactly what triggers the rule.

The classical rule appears in sections 323 and 324A of the Restatement (Second) of Torts (1965). One who undertakes to render services necessary for another’s protection becomes liable for negligent performance. Most states limit the duty to cases where negligent performance increased the risk or someone relied on the undertaking.

Courts already apply the rule to monitoring. In Sall v. T’s, Inc. (Kansas 2006), a golf course that undertook weather monitoring and warnings faced a triable undertaking claim after a lightning strike. The course owed no general duty to forecast weather. The decision to monitor created the duty to monitor with care.

Agents break the rule’s quiet assumption that undertakings are episodic. A deployer that builds crisis detection, fraud flags or output filters runs the undertaking continuously. Each alert is a performance, and each unanswered alert is a candidate for negligent performance. Users who see a safety feature rely on it, which satisfies the reliance branch.

The direction of the break expands deployer duty. The rule converts foresight into duty. Contingent watch condition: the first reasoned ruling that applies section 323 or 324A to an AI safety feature.

A deployer that builds monitoring takes on the duty to answer it, so response capacity matters as much as detection.

VI. Imputed Knowledge: What the Agent Knew

Imputed knowledge decides when a company knows what its systems capture. Agents make the doctrine urgent because they generate findings faster than people review them.

The classical rule imputes an employee’s knowledge to the employer when the employee acquired it within the scope of employment. United States v. Bank of New England (1st Cir. 1987) went further and aggregated what several employees knew into collective corporate knowledge. Courts have since met that doctrine with wide skepticism, and many reject collective knowledge as proof of intent.

Agents put knowledge where no person reads it. Evaluation results, flagged outputs and incident logs capture risks at machine speed. The open question is whether a log no person reviewed counts as company knowledge.

The direction of the break splits by purpose. For notice in negligence, imputation of an unread log is a modest step, because the company built the system that produced it. For intent in fraud or securities claims, courts that already resist collective knowledge will resist a log that no person ever held. Unread logs therefore expand negligence exposure far more than fraud exposure.

Imputed knowledge is the doctrine insurers should watch most closely. A court that imputes logs for notice hands carriers a known-loss or prior-knowledge argument from the same logs. Contingent watch condition: the first reasoned ruling that imputes an unreviewed automated finding to a company for notice.

Logs become notice before they become intent, and the gap between the two decides whether a claim sounds in negligence or fraud.

VII. Apparent Authority: Permissions as Authority

Apparent authority binds a principal to what others reasonably believe its actor may do. Agents hold permissions and present themselves to counterparties, which brings the doctrine into every deployment.

The classical rule binds a principal when a third party reasonably believes an actor is authorized, and that belief traces to the principal’s own manifestations. The Restatement (Third) of Agency (2006) frames apparent authority in section 2.03 as the power of “an agent or other actor.” The comment to section 1.04 treats computer programs as instrumentalities of the persons who use them, a position it qualifies with the words “at present.”

The combination already reaches AI agents. Software need not be a legal agent for apparent authority to bind the deployer, because section 2.03 covers other actors. The federal E-SIGN Act and state electronic-transactions statutes give contracts formed by electronic agents legal effect when the agent’s action is attributable to the person bound. Moffatt v. Air Canada (2024), a small-claims tribunal ruling worth $650.88, rejected the argument that a chatbot answered for itself.

Section 3.11 adds the timing rule. Ending actual authority does not by itself end apparent authority, which lasts until a third party can no longer reasonably believe the actor is authorized. A deployer that revokes an agent’s permissions but leaves its endpoint, persona or credentials visible to counterparties can remain bound.

The direction of the break binds the deployer to what its agent communicates. Granted permissions work like actual authority, and what counterparties see works like apparent authority. Contingent watch condition: the first U.S. ruling that binds a deployer to a transaction or representation by an AI agent on apparent-authority grounds.

A permission map becomes a liability map, and revocation protects the deployer only once counterparties know.

VIII. Shared Control: Res Ipsa Loquitur Meets the AI Stack

Shared control governs the inference a plaintiff draws when an injury speaks for itself. The AI stack splits control across layers, which tests whether that inference survives.

The classical rule lets a plaintiff infer negligence when an injury ordinarily does not occur without negligence and the instrument sat in the defendant’s exclusive control. Courts relaxed exclusive control for joint control, most famously in Ybarra v. Spangard (1944). Many courts still reject the inference when defendants held control in sequence rather than together.

An AI stack splits control in sequence. The developer trains the model, the integrator configures it and the deployer grants permissions and serves users. Each defendant can point to the next one, and the plaintiff often cannot identify which layer failed.

The direction of the break runs against plaintiffs. Opacity looks like the ideal case for res ipsa loquitur, yet consecutive control defeats the inference in many jurisdictions. Contingent watch condition: the first reasoned ruling that accepts or rejects res ipsa loquitur against more than one layer of an AI stack.

The doctrine plaintiffs expect to help them may fail, and the failure shifts weight toward contract allocation among the layers.

IX. Six Competing Duties Reward and Punish the Same Act

The four doctrines collide with each other, with the two gates and with the wider common law. Six competing duties govern deployer strategy, and each pairs a rule that rewards an act with a rule that punishes the same act.

Monitor versus answer. The Hand formula rewards monitoring, because detection lowers the expected loss a precaution must beat. The undertaking rule punishes monitoring that goes unanswered. A deployer that monitors more than it can answer moves exposure from the Hand formula to the undertaking rule.

Retain versus forget. Retained logs prove diligence and support a reasonable-care defense. The same logs may count as imputed knowledge and feed a known-loss argument in coverage. Deleting them after litigation becomes reasonably foreseeable invites sanctions under Federal Rule of Civil Procedure 37(e), so the only safe retention choice is the one set before any claim arises.

Revoke versus notify. Revoking an agent’s permissions ends actual authority. Apparent authority survives until counterparties can no longer reasonably believe the agent is authorized. Revocation without notice therefore protects the deployer internally and leaves it bound externally.

Shared control versus shared fault. Res ipsa loquitur may fail across a consecutive AI stack, which pushes plaintiffs to sue every layer. Joint and several liability and contribution then divide the loss, and vendor contracts decide the final split. The doctrine that narrows the plaintiff’s proof route widens the defendants’ fight with each other.

Foresee versus intervene. Superseding cause rewards a defendant that can point to an independent act between its conduct and the harm. Foreseeability of type punishes the same defendant when its own testing documented that type. Pleading an agent’s autonomy or a user’s misuse therefore works only for harm types the defendant never found.

Tort versus coverage. The foreseeability paradox closes the set. The same dated finding proves notice in tort and supports a known-loss defense in coverage. A finding answered before binding reads as care in both forums.

Answerable monitoring resolves three of the six directly: monitor versus answer, foresee versus intervene and tort versus coverage. The other three need their own instruments: a retention schedule set before any claim, counterparty notice at revocation and control allocated by contract.

Answerable monitoring has four parts: detection, escalation to a named owner, containment and verification that the containment worked. Closing a ticket without verified containment is weak evidence of care. The observable test: every alert class has a named owner, a response window and a verification step.

The competing duties leave deployers four instruments: answerable monitoring, a retention schedule, counterparty notice and contract allocation. Carriers can check for all four at renewal.

X. Standards, Loss Rules, Coverage Doctrines and Statutes Raise the Cost of Every Move

The four doctrines sit inside a wider structure of standards, loss-routing rules and coverage doctrines.

The supporting doctrines fall into four groups: those that set the standard, those that route the loss, those that read foresight backward in coverage and those that name duties courts have not yet resolved.

Doctrines that set the standard. The Hand formula weighs the burden of a precaution against the probability and size of the loss. The T.J. Hooper (1932) adds that industry custom does not set the standard, because a whole industry can lag reasonable prudence. Negligence per se lets a court borrow a statute such as a state AI law as the standard of care.

Design defect asks whether a reasonable alternative design existed, under section 2 of the Restatement (Third) of Torts: Products Liability (1998). Section 10 imposes a duty to warn after sale only when the seller knows of a substantial risk and can identify the users to warn. Section 11 recognizes a recall duty only when a government orders one or the seller performs a voluntary recall negligently.

Several states reject post-sale duties outright. A developer that voluntarily rolls back a model takes on a duty to execute the rollback with reasonable care. The voluntary fix creates its own standard, which links section 11 to the undertaking rule in Section V.

Doctrines that route the loss. The economic loss rule keeps many purely financial losses out of tort and pushes them into contract. Agent losses such as a misdirected payment therefore turn on vendor terms more than on negligence. Subrogation hands the paying carrier the insured’s rights, along with every fee cap and waiver the insured signed.

Doctrines that read foresight backward. Known-loss and prior-knowledge provisions ask whether the loss was in progress or known at binding. Expected or intended injury provisions ask whether the insured foresaw the harm with something close to certainty. Misrepresentation and rescission turn questionnaire answers into grounds to void a policy. The duty to defend and construction against the drafter push the other way, so new AI exclusions will be litigated on their wording.

Number of occurrences decides correlated loss. Courts split between counting causes and counting effects. One model failure across many insureds can read as one occurrence or many, and the answer sets limits, retentions and reinsurance recovery. Who Absorbs Correlated AI Loss sets out the aggregation stakes behind that count.

Candidate duties courts have not yet resolved. Plaintiffs plead affirming, engagement-oriented chatbot design as a defect, as in Raine. Others plead harms to minors from companion models. A duty to act on detected crisis signals draws on the undertaking rule in Section V, and fraud executed by an agent with delegated authority draws on the apparent-authority rule in Section VII.

Statutes that move the doctrines. Statutes act on the same gates and doctrines from outside the common law. Each one below shifts a doctrine in a known direction, and none decides who finally pays.

Colorado has written attribution by conduct into statute. SB 26-189 allocates liability between developers and deployers by comparative fault, which narrows the room vendor terms leave for allocation by contract. The AI LEAD Act would go further only if enacted: it would void unconscionable liability caps and let a deployer stand in for an absent or insolvent developer.

The supporting structure explains why each seat moves the way it does, because every move carries a doctrine or statute that makes it costly.

XI. Carriers Test the Competing Duties at Renewal Before Courts Rule

Carriers apply the common law before courts do. Renewal makes carriers the first institution to test the competing duties, and the market splits into segments that move differently.

Every policy renews on its own date, and each renewal turns the insured’s inquiry file into an underwriting exhibit. The coverage contest is moving from whether a policy excludes AI to what the insured knew and when. January 1, 2027 marks the largest single cohort, not the whole market.

The six competing duties each surface in the application. Monitoring questions test monitor versus answer, and retention questions test retain versus forget. Questions on agent permissions and decommissioning test revoke versus notify, and vendor contracts test shared control versus shared fault. Testing and red-team questions test foresee versus intervene.

The market splits into three segments. Segment A covers admitted small and mid-size commercial general liability on standard forms, where Verisk exclusions attach most easily. Segment B covers specialty errors and omissions and technology lines, where affirmative cover conditioned on evidence is most plausible. Segment C covers large accounts on negotiated wording, where silent cover is most likely to persist.

The cyber market supplies the routine framing. Cyber moved from silent cover to exclusions to specialist standalone policies, and the NAIC report shows surplus-lines carriers now writing nearly two-thirds of cyber premium. AI liability is following the same sequence at an earlier stage. Who Writes AI Insurance Coverage Conditions traces the exclusion filings and coverage conditions that drive the segment split.

The renewal questionnaire is the first institution to test whether a deployer resolved the competing duties, and each segment answers on its own terms.

XII. Attribution Follows Conduct Through Four Stages

Attribution by conduct replaces attribution by role. Developer and deployer can both be exposed in one case, and being named does not decide who pays.

Exposure attaches to conduct, not to a party’s place in the stack. Deployer-controlled conduct covers permissions, representations and the response to detected risk. Developer-controlled conduct covers design, warnings and upstream changes. One injury can implicate both sets, so a plaintiff can reach both parties.

MacPherson keeps the developer’s duty alive past the intermediary when the danger was foreseeable. Perplexity shows one statutory context where direction moved attribution to the user. The Ninth Circuit limited its holding to computer access and reserved tort, so whether direction moves tort attribution to deployers remains a hypothesis.

Exposure then passes through four stages. A deployer is named in a complaint, and the claim survives a liability challenge. Coverage responds or fails. The deployer then recovers from another responsible party or keeps the loss.

Each stage needs its own evidence, and a forecast about one stage cannot establish the next. A filed complaint does not show liability. A liability finding does not show retained loss, because a tender seeks to move loss upstream, and accepted indemnity or recovery determines whether it succeeds.

A forecast about who gets named therefore says nothing about who finally pays.

XIII. Renewal and Litigation Run on Independent Clocks

Two clocks move a deployer’s exposure. Crossing coverage with claim status shows where each deployer stands, and the order in which the clocks arrive stays open.

The renewal clock moves coverage by policy date and segment. The litigation clock moves claims by filing, ruling and resolution. The two clocks run independently, so a deployer’s cover can change before any claim arrives, or a claim can arrive under wording written years earlier.

Recovery runs across all four cells. A named deployer’s retained loss depends on indemnity terms, liability caps and the developer’s solvency. A coverage action may also arise under older wording, so new exclusions and coverage litigation need separate causal tests.

Which clock reaches a given segment first is the decisive open question. Hypothesis H12 tests the ordering in Segment A. The paper states no ordering as a finding.

The matrix turns a doctrinal map into a decision tool, because each cell carries a different exposure and a different response.

XIV. Every Stakeholder Moves One Stage, and Only Deployers Sit in All Four

The stakeholder board names the institutions whose choices move deployers through the four stages. Each stakeholder holds a stage, an objective, a binding constraint and a move that would change the game.

MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through MindCast AI Proprietary Cognitive Digital Twin Foresight Simulations (MP CDT FS). Behavioral Economics supplies the decision rules. Game Theory supplies the payoff structure. Predictive simulations emerge from the combination.

The game evolution MindCast foresees runs through the inquiry file. Renewal forms, questionnaires and claims files will turn on the same dated findings that tort plaintiffs plead. Behavioral Economics explains why: underwriters anchor on the nearest written standard when the technical file is unreadable. Game Theory explains the result: the insured chooses what to disclose while two readers interpret the same disclosure against it.

The persistence path stays live. Carriers may renew terms unchanged, and courts may resolve early agent cases under ordinary negligence. The competing duties then stay latent until the first large claim.

Every stakeholder moves one stage at least, and only deployers sit in all four.

XV. Carriers Hold Three Seats While Lawmakers and Enforcers Set the Pace

Stakeholder positions differ because each group moves a different stage of the chain. Each group below carries the mechanism that drives it and the Snapshot Predictions that test it.

Carriers as Insurers and Defendants

Carriers defend on two fronts. As insurers they face coverage actions over agentic losses, and as AI users they face bad-faith and contract claims over their own automated claim decisions. A federal court in Minnesota ordered broad discovery into an insurer’s AI-driven coverage denials in Estate of Lokken v. UnitedHealth Group in March 2026. Policyholders are now building bad-faith theories on the same pattern.

Game Theory explains the bind. Exclusions that cut silent exposure also push insureds toward coverage litigation, and every automated claim decision becomes discoverable in the next bad-faith suit. Behavioral Economics adds that underwriters anchor on the first adverse coverage ruling, so one decision resets terms across a whole segment. Snapshot Predictions P7 · S4 · P9–P11 · S5 · P14 · P19 test the position.

Carriers as Plaintiffs

Carriers also sue. Nippon Life Insurance Company of America sued OpenAI directly in March 2026, alleging that ChatGPT practiced law without a license. In cyber, Ace American Insurance Company brought a 2025 subrogation actionagainst its insured’s technology vendors after a ransomware loss, and the template transfers to AI vendors.

Subrogation inherits every term the insured signed. A New York federal court held in July 2026 that an insurer had no enforceable subrogation rights against a vehicle-valuation software maker after paying a class-action loss. Vendor fee caps and waivers in a deployer’s AI contracts therefore cap the carrier’s recovery too. P18 tests the position.

Frontier AI Firms

Frontier AI firms carry exposure through two channels. Directly, they defend consumer suits and attorney general actions as self-deployers, as OpenAI does in Raine and in the October 14 state actions. Indirectly, enterprise deployers that get named will tender to them, and carriers that pay will seek subrogation.

Contract terms decide the indirect channel. Fee caps and narrow indemnities hold exposure down until a court, regulator or statute overrides them. The proposed AI LEAD Act would limit a developer’s strict liability where a deployer made a major unauthorized change, which would write attribution by conduct into federal law. Snapshot Predictions P1 · P2 · P5 · S3 · P12 test the position.

Other AI Firms

Other AI firms build applications, agents and integrations on frontier models. Character Technologies, Perplexity and Workday each show a different exposure: insolvency under litigation pressure, attribution of agent conduct and treatment of screening software as an employer’s agent in Mobley v. Workday.

Thin balance sheets change the game. A plaintiff facing an undercapitalized defendant looks for solvent partners and insurance, as the Character bankruptcy shows. Insolvency can therefore work as a liability exit for the firm and a liability transfer to its partners. P3, P4 and P20 test the position.

State Lawmakers

State lawmakers write the duties the other stakeholders litigate. California removed the autonomy defense through section 1714.46 and regulated companion chatbots through SB 243, both from January 1, 2026. Colorado’s 2024 AI Act drew a federal-court stay in April 2026 after xAI sued and the Department of Justice intervened. Colorado then repealed and replaced it with SB 26-189, which allocates fault between developers and deployers from January 1, 2027.

The December 2025 executive order exempts child-safety laws only from its proposed federal preemption statute. The Department of Justice task force it created carries no such exemption, so child-safety framing narrows the legislative threat without shielding a state law from suit. P16 and P21 test the position.

State Attorneys General

State attorneys general move fastest of any public actor. Kentucky filed the first state action against an AI chatbot in January 2026, Florida followed in June and eight states filed on October 14. Texas filed its own suit on October 23. Each office files in its own court under its own consumer law, which multiplies forums without a federal order.

Bankruptcy tests their reach. The police-power exception may keep state enforcement moving against Character Technologies while private suits stay paused. The timing edge favors attorneys general over private plaintiffs. P2, S1 and P5 test the position.

Federal Lawmakers and the Executive Branch

Federal lawmakers hold two levers that pull in opposite directions. Senators Dick Durbin and Josh Hawley introduced the AI LEAD Act on September 29, 2025. The bill would classify AI systems as products and create a federal product-liability cause of action. The December 11, 2025 executive order pushes the other way through a Department of Justice AI Litigation Task Force that challenges state AI laws.

Neither lever has resolved. The bill has not been enacted, and the task force’s first courtroom result came through intervention in a private suit against Colorado. A federal product-liability statute would itself displace parts of state law, so expansion and preemption can arrive together. P15 and P16 test the position.

Consumer Groups

Consumer groups set the agenda the enforcers follow. A coalition led by the Consumer Federation of America asked the Federal Trade Commission in June 2025 to investigate therapy chatbots on Character.AI and Meta. The agency opened a study of companion chatbots under section 6(b) of the FTC Act in September 2025. Fairplay and the Transparency Coalition backed the AI LEAD Act.

Their published tests carry legal weight. A documented test showing a harmful chatbot response becomes evidence of a foreseeable type of harm. The same evidence feeds notice in tort and known-loss arguments in coverage, so advocacy research enters the two gates and the tort-versus-coverage conflict directly. P13 and P17 test the position.

Each stakeholder moves a different stage, and several sit in more than one seat. Carriers alone appear as insurer, defendant and plaintiff, which makes the deployer’s vendor contract the one document that governs all three positions.

XVI. Legal Incidents Cluster on Self-Deploying Developers

The incident baseline sets the starting rates the Snapshot Predictions move against. Two incident streams carry the dependent variables. Consumer incidents are the harms agents and chatbots cause. Legal incidents are the complaints, enforcement actions, rulings and coverage disputes those harms produce.

Consumer incidents run near 50 a month in aggregate public counts. The AI Incident Database added 148 incident entries between May and July 2026. The aggregate count does not isolate agentic harm. The conversion test in P8 therefore uses a fixed agentic-harm category, deduplicated by incident and matched to the legal-incident window.

Legal incidents in 2026 arrived in a clear sequence. Each one engages a doctrine or a competing duty mapped in Sections IV through IX.

The harm-driven legal incidents share one feature. Each targets a developer that runs its own consumer product, so developer and deployer sit in one seat. No public complaint yet charges an independent enterprise deployer with an agent’s conduct under the four doctrines.

Coverage incidents remain sparse. Carriers have filed generative-AI exclusions at volume during 2026. One businessowners carrier sought a declaratory judgment in Illinois federal court over biometric-privacy claims arising from AI-handled phone orders.

Legal incidents cluster on self-deploying developers, and coverage disputes over agentic loss sit at zero.

XVII. Six Rival Accounts Predict Different Paths

Competing explanations predict a different path, and each names the observation that would favor it. All six stay live until the first rulings and renewals arrive.

• Courts stay with ordinary negligence. Early agent cases resolve under the Hand formula without reaching the four doctrines. Observation favoring it: reasoned rulings that decide agent claims on breach alone.

• Contracts resolve allocation first. Vendor terms and indemnities divide loss before any doctrine matters. Observation favoring it: dismissals on contractual grounds in multi-party AI disputes.

• Legislation displaces state doctrine. A statute replaces common-law rules for AI deployment before courts apply them. Observation favoring it: an enacted statute that sets agent liability rules directly.

• Liability stays with developers. Complaints keep naming only the model builder. Observation favoring it: P6 resolves false by March 31, 2027.

• The clocks run in reverse. Claims reach some segments before any renewal changes their cover. Observation favoring it: S5 resolves false in the Segment A sample.

• Federal preemption resets the board. Congress preempts state AI liability rules, or federal courts enjoin them. Observation favoring it: by March 31, 2027, an enacted preemption statute or a federal injunction or stipulated stay against a state AI statute outside Colorado.

Each rival account has a named observable, so the first rulings, renewals and deployer complaints will separate them.

XVIII. MindCast Simulation Predictions: Complaints and Enforcement Lead, Carrier Litigation Trails

The Snapshot carries 26 MindCast Foresight Simulation Predictions, called Snapshot Predictions below, across 20 hypotheses: 21 Primary Simulation Predictions marked P and 5 Secondary Simulation Predictions marked S. A secondary prediction states its condition first and resolves only once that condition occurs. Nineteen carry bands, and seven stay pending until their samples exist.

Each band gives a central estimate with its uncertainty range. Dated predictions resolve false when the event does not occur in the window, and each resolves against its named verification source.

Segment predictions resolve within a renewal sample of policies that renew between December 1, 2026 and March 31, 2027. P8 waits for a fixed denominator of agentic consumer incidents.

P1. Plaintiffs file a new complaint by January 31, 2027 alleging that an AI developer or deployer detected a user’s crisis or threat signal and failed to act (85%; range 75–92%). Tests H1 (monitor versus answer). Falsifier: no qualifying complaint by January 31, 2027. Verification source: federal and state dockets.

P2. A state attorney general files a new complaint by March 31, 2027 alleging that an AI company concealed internal safety findings (55%; range 40–70%). Tests H2 (imputed knowledge). Falsifier: no qualifying complaint by March 31, 2027. Verification source: state attorney general releases.

P3. A court cites Amazon v. Perplexity by March 31, 2027 to attribute an AI agent’s act in a claim outside the computer-access statutes (20%; range 10–30%). Tests H3 (reach of Perplexity). Falsifier: no qualifying citation by March 31, 2027. Verification source: published opinions.

P4. A Character.AI claimant takes a documented step by January 31, 2027 against a non-debtor’s own conduct or available insurance (65%; range 50–80%). Tests H4 (collectibility redirects claims). Falsifier: no qualifying motion or pleading by January 31, 2027. Verification source: bankruptcy docket No. 26-11603 and related dockets.

P5. At least three states beyond the ten with actions at the cutoff file their own actions against OpenAI by March 31, 2027 (50%; range 35–65%). Tests H6 (multistate herding). Falsifier: fewer than three new state actions by March 31, 2027. Verification source: state attorney general releases.

P6. A public complaint names an independent enterprise deployer by March 31, 2027 on a tort or agency theory for an AI agent’s action or statement (45%; range 30–60%). Tests H7 (deployer migration). Falsifier: no qualifying complaint by March 31, 2027. Verification source: federal and state dockets.

P7. A carrier files a public coverage action by March 31, 2027 over a loss caused by an AI agent’s action (15%; range 5–25%). Tests H8 (tort versus coverage). Falsifier: no qualifying coverage action by March 31, 2027. Verification source: federal and state dockets.

P8. The ratio of new AI legal incidents to new agentic consumer incidents in the first quarter of 2027 exceeds the third-quarter 2026 ratio (band pending a fixed incident denominator). Tests H9 (conversion velocity). Falsifier: the first-quarter 2027 ratio equals or falls below the third-quarter 2026 ratio. Verification source: AI Incident Database, OECD AI Incidents Monitor and dockets.

P9. Carriers attach a generative-AI exclusion to more than half the policies in the Segment A sample (band pending the renewal sample). Tests H10 (coverage by segment). Falsifier: half or fewer of the sampled policies carry the exclusion. Verification source: endorsement schedules in the sample.

P10. At least one Segment B carrier in the sample offers affirmative agentic cover conditioned on inquiry evidence (band pending the renewal sample). Tests H10 (coverage by segment). Falsifier: no Segment B carrier in the sample offers such cover. Verification source: policy forms and quotes in the sample.

P11. More than half the Segment C renewals in the sample carry no generative-AI exclusion (band pending the renewal sample). Tests H10 (coverage by segment). Falsifier: half or more of the Segment C renewals carry an exclusion. Verification source: endorsement schedules in the sample.

P12. A court issues a written ruling by March 31, 2027 on the merits of a user-misuse or circumvention defense in an AI chatbot harm case (20%; range 10–30%). Tests H11 (foresee versus intervene). Falsifier: no merits ruling by March 31, 2027. Verification source: federal and state dockets.

P13. A new pleading, motion or ruling filed by March 31, 2027 contests whether a post-detection or post-suit safety measure addressed the risk (60%; range 45–75%). Tests H13 (response adequacy in litigation). Falsifier: no qualifying filing after November 2, 2026 and by March 31, 2027. Verification source: pleadings, motions and rulings.

P14. An underwriting document in the Segment B sample asks for remediation evidence beyond monitoring capability by March 31, 2027 (band pending the renewal sample). Tests H13 (response adequacy in underwriting). Falsifier: no such request in the Segment B sample. Verification source: dated questionnaires and coverage conditions.

P15. The Senate Judiciary Committee marks up the AI LEAD Act or a successor bill that keeps its federal product-liability claim by March 31, 2027 (8%; range 3–15%). Tests H14 (federal liability legislation). Falsifier: no qualifying markup by March 31, 2027. Verification source: Congress.gov committee actions.

P16. The Department of Justice files or joins a challenge by March 31, 2027 to a state AI statute outside Colorado (45%; range 30–60%). Tests H15 (preemption by litigation). Falsifier: no qualifying filing by March 31, 2027. Verification source: Department of Justice releases and federal dockets.

P17. A consumer or child-safety coalition files a Federal Trade Commission complaint or amicus brief by March 31, 2027 targeting an agentic AI product rather than a companion chatbot (35%; range 20–50%). Tests H16 (advocacy follows the agents). Falsifier: no qualifying filing by March 31, 2027. Verification source: coalition releases and court dockets.

P18. A carrier files a subrogation action by March 31, 2027 against an AI developer or vendor after paying an insured’s AI-related loss (12%; range 5–20%). Tests H17 (carriers as plaintiffs). Falsifier: no qualifying action by March 31, 2027. Verification source: federal and state dockets.

P19. A new complaint outside health insurance alleges by March 31, 2027 that a carrier’s AI tool wrongly denied or undervalued a claim (45%; range 30–60%). Tests H18 (carriers as AI deployers). Falsifier: no qualifying complaint by March 31, 2027. Verification source: federal and state dockets.

P20. An AI application or agent company other than Character Technologies files for bankruptcy or documents a wind-down by March 31, 2027 while facing harm litigation (25%; range 15–40%). Tests H19 (insolvency as liability exit). Falsifier: no qualifying filing or notice by March 31, 2027. Verification source: bankruptcy dockets, state dissolution filings and company notices.

P21. A state enacts a statute by March 31, 2027 assigning deployers duties for AI agent actions beyond disclosure, with Colorado’s SB 26-189 excluded (20%; range 10–30%). Tests H20 (state deployer duties). Falsifier: no qualifying enactment by March 31, 2027. Verification source: state session laws.

S1. Given a contested motion, the bankruptcy court treats a state enforcement action against Character Technologies as excepted from the stay by March 31, 2027 (65%; range 50–80%). Tests H5 (public enforcement during a stay). Falsifier: the court applies the stay to state enforcement on a contested motion. Verification source: bankruptcy docket No. 26-11603.

S2. Given P6, the complaint alleges the deployer’s own permission, representation or response failure (70%; range 55–85%). Tests H7 (deployer migration). Falsifier: the qualifying complaint rests only on upstream design. Verification source: the complaint.

S3. Given P6, the named deployer takes a public upstream recovery step within 90 days of service (40%; range 25–55%). Tests H7 (deployer migration). Falsifier: no third-party complaint, cross-claim or filing that describes a tender within 90 days. Verification source: dockets and public filings.

S4. Given P9, a carrier files a public coverage action by March 31, 2027 over an agentic loss (band pending the renewal sample). Tests H8 (tort versus coverage). Falsifier: no qualifying coverage action after P9 resolves true. Verification source: federal and state dockets.

S5. Given at least one Segment A sample deployer with both an AI exclusion and an agent claim in the window, the exclusion attaches before the claim for more than half of them (band pending the renewal sample). Tests H12 (clock ordering). Falsifier: the claim arrived first for half or more of them. Verification source: endorsement schedules matched to dockets.

The definitions fix what counts. P5 excludes the ten states with actions at the cutoff: Florida, the eight October 14 filers and Texas. S3 counts a third-party complaint, a cross-claim or a filing that attaches or describes a tender. P12 counts merits rulings only, and P20 counts an assignment for creditors, a dissolution or a public shutdown notice.

P6 counts negligence, voluntary undertaking, apparent authority and misrepresentation. Wiretap and other privacy-interception claims fall outside P6, because they already name deployers.

P15 spans two calendars. The rest of the 119th Congress runs to January 3, 2027 on a crowded post-election schedule. In the 120th Congress the bill needs reintroduction and the committee must reorganize before any markup.

P6 carries the most weight, because it tests whether claims reach the enterprise deployer, where the four doctrines point. S2 and S3 then test the pleaded theory and the recovery route. A liability finding or retained loss lies beyond this window.

Plaintiff and enforcement moves rank highest. Crisis-signal complaints (P1) lead, with the stay exception (S1), claims against Character’s partners (P4) and new contests over safety measures (P13) close behind. A public coverage action (P7) and a subrogation suit (P18) both fall below one in five.

Public legal activity concentrates on complaints and enforcement. Renewal changes may advance without producing public coverage litigation.

XIX. Six Events Change the Board

Each event in the Snapshot changes the board for the next hypothesis. Each event below carries five entries, a class and a validity condition.

The class separates three cases. Continuation updates information inside the current game. Equilibrium migration changes which outcome persists. Game mutation changes who plays, which moves exist or which rules govern, and it is the exception to ordinary game evolution.

Event 1. Character Technologies Chapter 11, October 12, 2026. Class: game mutation.

• What changes: players and rules. A bankruptcy court and a buyer enter, and the automatic stay pauses private suits against the debtor, which remains an actor.

• Paths opened and closed: stay-relief motions, creditor claims and claims against non-debtor partners open. Direct private discovery against the debtor closes while the stay runs.

• Positions: private plaintiffs convert into creditors. State enforcement may survive under the police-power exception.

• Moves gained and lost: claimants gain routes to non-debtor conduct and insurance. The debtor loses control of the litigation calendar.

• Conditional Snapshot Predictions: P4 and S1 activate.

• Validity condition: S1 lapses if the bankruptcy case ends before any stay ruling.

Event 2. The renewal cohorts from December 2026 through March 2027. Class: equilibrium migration in segments where exclusions or conditioned cover take hold, and continuation where terms hold.

• What changes: payoffs and information. Coverage terms reset by policy date, and application answers become underwriting evidence.

• Paths opened and closed: where Segment A exclusions take hold, silent cover for agent losses closes. Where Segment B carriers condition cover on evidence, evidence-based cover opens. Where Segment C wording holds, current routes stay open.

• Positions: deployers with answered and verified findings survive on every branch. Deployers with open findings move to the excluded column wherever an exclusion attaches.

• Moves gained and lost: carriers gain the questionnaire as a screening move. Deployers lose the option to stay silent about findings.

• Conditional Snapshot Predictions: S4 activates when P9 resolves true. S5 resolves on the Segment A sample.

• Validity condition: P7 stays valid in every segment.

Event 3. The first complaint against an independent enterprise deployer. Class: equilibrium migration.

• What changes: players. Enterprise deployers become defendants in their own right.

• Paths opened and closed: undertaking, imputed-knowledge and apparent-authority counts against deployers open. The defense that only the model builder answers faces its first test.

• Positions: vendor indemnities convert from contract terms into live recovery routes. Deployers without contract allocation lose their clearest route upstream.

• Moves gained and lost: deployers gain tenders and third-party claims against developers. Developers lose distance from deployer conduct.

• Conditional Snapshot Predictions: P6 resolves true, and S2 and S3 activate.

• Validity condition: P1 through P3 stay valid, because the event adds defendants without changing doctrine.

Event 4. The first reasoned ruling on one of the four doctrines. Class: continuation if the court decides on ordinary negligence, and equilibrium migration if it applies the doctrine to an agent.

• What changes: rules. One jurisdiction fixes the doctrine for later plaintiffs there.

• Paths opened and closed: the accepted theory opens for borrowing. The rejected theory closes in that jurisdiction.

• Positions: pleadings built on the accepted doctrine survive. Pleadings built on the rejected doctrine convert or strand.

• Moves gained and lost: plaintiffs gain a cited basis. Defendants lose the novelty argument.

• Conditional Snapshot Predictions: the matching hypothesis converts into a dated fork in the next Snapshot.

• Validity condition: P3 resolves only on a ruling outside the computer-access statutes.

Event 5. The first public coverage action over an agentic loss. Class: equilibrium migration.

• What changes: information. The first coverage pleading shows which defense carriers choose and under which wording.

• Paths opened and closed: known-loss and expected-injury defenses enter litigation. The assumption that exclusions alone resolve agent losses faces its first test.

• Positions: the filing exposes the competing positions. Insureds with dated and answered findings hold the stronger response, and insureds with open findings face the known-loss defense.

• Moves gained and lost: carriers gain a template defense. Insureds gain construction-against-drafter arguments on new wording.

• Conditional Snapshot Predictions: P7 resolves true, and S4 resolves on the Segment A outcome.

• Validity condition: P9 through P11 stay valid, because the action does not change the sample’s issued wording.

Event 6. The Department of Justice intervention in xAI’s challenge to the Colorado AI Act, April 24, 2026. Class: game mutation.

• What changes: players. The federal government enters a private challenge to a state AI statute.

• Paths opened and closed: preemption by litigation opens without a federal statute. Unchallenged enforcement of broad state AI duties closes in that case.

• Positions: child-safety framing narrows the proposed preemption statute but not the task force’s suits. Broad algorithmic-duty laws convert into litigation targets, and Colorado answers by repealing and replacing its act.

• Moves gained and lost: AI firms gain a federal ally against state duties. States lose the presumption that their statutes take effect on schedule.

• Conditional Snapshot Predictions: P16 and P21 become live.

• Validity condition: P21 resolves on enactment. A later injunction counts as a subsequent game transition.

Once the first coverage action and the first doctrinal ruling fix the rules, the coverage path becomes stable enough for longer forecasts.

Each event carries its own map of what lies past it, so each stakeholder can plan for a branch before it opens.

XX. Risk Mitigation: Moves Each Stakeholder Controls

Stakeholder strategy depends on which stage a party moves and which seat it occupies.

The four instruments from the competing duties hold in every cell of the coverage and claim matrix. Each entry pairs one or two Snapshot Predictions with one stakeholder and states severity and probability on separate axes. Entries describe exposure and options, not legal, investment or fiduciary advice.

Entry 1. 💼 Executives at Segment A deployers, S5.

• Prediction and band: exclusions attach before claims for most Segment A deployers that receive both an exclusion and a claim (band pending the renewal sample). Severity: high. Probability: pending.

• Exposure: latent agent exposure with no cover for the excluded conduct, measured in quarters of operating budget.

• Mitigating actions: the risk function requests the full endorsement schedule 60 days before each renewal date. The finance function sizes a retained-loss reserve before binding. Procurement obtains a quote for specialty affirmative cover against the reserve before the renewal date.

• Residual exposure: losses from agents the company does not know it runs stay outside every schedule.

Entry 2. 💼 Deployers named in an agent complaint, P6 and S3.

• Prediction and band: a first complaint names an independent enterprise deployer (45%; range 30–60%). Severity: high. Probability: moderate.

• Exposure: discovery scope covering every open finding and unreviewed log, plus defense spend that indemnity may or may not return.

• Mitigating actions: the risk function assigns an owner, a response window and a verification step to each alert class before the next renewal. Legal sets a log-retention schedule by January 31, 2027. Procurement negotiates indemnity carve-outs from vendor fee caps before the next contract renewal.

• Residual exposure: findings closed after the first complaint still carry their original dates.

• Companion link: the alert-ownership action also reduces exposure under the incident-response theory in FTC Investigation of Anthropic and OpenAI, and the credit is claimed here once.

Entry 3. 💼 Deployers decommissioning agents, P3.

• Prediction and band: a court cites Amazon v. Perplexity to attribute an AI agent’s act outside the computer-access statutes (20%; range 10–30%). Severity: medium. Probability: low.

• Exposure: transactions bound by decommissioned agents, measured in counterparty contracts still reachable through retired endpoints or credentials.

• Mitigating actions: engineering retires agent endpoints and credentials at revocation. The commercial team sends termination notice to counterparties that dealt with the agent before March 31, 2027.

• Residual exposure: counterparties who learned of the agent indirectly may still claim reasonable reliance.

Entry 4. 💼 Carriers as insurers, P9 through P11 and P7.

• Prediction and band: segment-level coverage paths remain pending the renewal sample, and a first agentic coverage action carries 15%; range 5–25%. Severity: high. Probability: low.

• Exposure: defense spend on the first coverage action plus silent exposure measured in aggregate limits per model vendor.

• Mitigating actions: underwriting adds questions on alert ownership, verification and agent decommissioning at each renewal in the sample window. The claims function reviews reservation-of-rights wording before the first agentic notice. Reinsurance buyers model one correlated loss scenario before treaty renewal.

• Constraint: each carrier decides its wording unilaterally and draws only on filed standard forms, which keeps the move clear of competitor coordination.

• Residual exposure: number-of-occurrences disputes remain open until a court rules.

Entry 5. 💼 Carriers as plaintiffs, P18.

• Prediction and band: a carrier files a subrogation action against an AI developer or vendor (12%; range 5–20%). Severity: medium. Probability: low.

• Exposure: paid AI-related losses the carrier cannot recover, measured in basis points of loss ratio on the affected book.

• Mitigating actions: underwriting requests each insured’s AI vendor contracts at renewal. The claims function sends subrogation notice within 30 days of each AI-related claim. Product teams offer a renewal credit for vendor contracts that preserve subrogation rights.

• Residual exposure: caps and waivers signed before the policy incepted still bind the carrier.

Entry 6. 💼 Carriers as AI deployers, P19.

• Prediction and band: a complaint outside health insurance alleges that a carrier’s AI tool wrongly denied or undervalued a claim (45%; range 30–60%). Severity: medium. Probability: moderate.

• Exposure: discovery scope over every automated claim decision in the affected line.

• Mitigating actions: the claims function attaches a dated human-review entry to each AI-assisted denial by the next quarterly audit. Compliance maps each AI claims tool to state unfair-claims-practices rules by January 31, 2027.

• Residual exposure: denials already issued stay discoverable.

Entry 7. 💼 Frontier AI firms, S3.

• Prediction and band: given P6, the named deployer takes a public upstream recovery step within 90 days of service (40%; range 25–55%). Severity: high. Probability: moderate.

• Exposure: indemnity demands from enterprise customers, measured in contract value under tender.

• Mitigating actions: legal adds a conduct-based allocation clause covering permissions, representations and response to enterprise terms at the next contract cycle. Safety teams attach a dated remediation entry to each evaluation finding before release.

• Residual exposure: consumer suits and attorney general actions reach the firm directly, whatever the enterprise terms say.

Entry 8. 💼 Other AI firms, P20 and P4.

• Prediction and band: another AI application or agent company enters bankruptcy or winds down while facing harm litigation (25%; range 15–40%). Severity: high. Probability: low.

• Exposure: claims that outlast the firm, measured in quarters of runway against the litigation reserve.

• Mitigating actions: finance buys affirmative AI liability cover sized to the litigation reserve before the next funding round. Legal negotiates allocation terms with each frontier model provider before the next renewal.

• Residual exposure: claims may follow assets to an acquirer.

Entry 9. 🏛️ State lawmakers, P21 and P16.

• Prediction and band: a state enacts deployer duties for AI agent actions beyond Colorado’s SB 26-189 (20%; range 10–30%). The Department of Justice files or joins a challenge outside Colorado (45%; range 30–60%). Severity: medium. Probability: low for enactment and moderate for a challenge.

• Exposure: session calendar spent on statutes that federal litigation stays or forces into replacement, as Colorado’s did within five weeks of xAI’s April 9 filing.

• Mitigating actions: legislative counsel drafts deployer duties against the six competing duties before the 2027 sessions open in January. Drafters place child-safety provisions in severable sections before introduction.

• Residual exposure: a federal statute may still preempt state duties.

Entry 10. 🏛️ State attorneys general, S1 and P5.

• Prediction and band: given a contested motion, the bankruptcy court treats state enforcement as excepted from the stay (65%; range 50–80%). Three more states sue OpenAI (50%; range 35–65%). Severity: medium. Probability: moderate to high.

• Exposure: enforcement calendar lost to the stay, measured in months of paused discovery.

• Mitigating actions: litigation divisions file police-power positions in the bankruptcy before the sale hearing. Offices that have not filed assess deployer targets under existing consumer law by March 31, 2027.

• Residual exposure: removal and consolidation motions may slow state-court actions.

Entry 11. 🏛️ Federal lawmakers, P15 and P16.

• Prediction and band: the Senate Judiciary Committee marks up the AI LEAD Act or a successor bill (8%; range 3–15%). The Department of Justice files or joins a challenge outside Colorado (45%; range 30–60%). Severity: medium. Probability: low for a markup and moderate for a challenge.

• Exposure: committee calendar spent while courts set allocation case by case.

• Mitigating actions: committee staff align the bill’s developer and deployer definitions with conduct-based attribution before any markup. Staff schedule a hearing on deployer duties and coverage gaps before March 31, 2027.

• Residual exposure: preemption litigation may moot parts of any bill.

Entry 12. 🛡️ Consumer groups, P17 and P13.

• Prediction and band: a coalition targets an agentic product (35%; range 20–50%). A new filing contests whether a safety measure worked (60%; range 45–75%). Severity: medium. Probability: moderate.

• Exposure: the filing window before renewal terms and state rules set, measured in comment periods and session calendars.

• Mitigating actions: research staff publish dated agent tests with a follow-up test of each vendor’s response in the same cycle. Legal staff prepare an agentic-product complaint before state sessions open in January.

• Residual exposure: firms may contest the test methodology.

Entry 13. ⚖️ Counsel for non-debtor co-defendants, P4.

• Prediction and band: a Character claimant moves against a non-debtor’s conduct or insurance (65%; range 50–80%). Severity: medium. Probability: moderate to high.

• Exposure: discovery scope and defense spend shifting onto licensing partners while the debtor sits stayed.

• Mitigating actions: the general counsel of each licensing partner reviews indemnity and license terms with the debtor before the sale closes. Litigation counsel prepares the response to stay-relief motions by January 31, 2027.

• Residual exposure: claims may follow assets transferred to a buyer.

Entry 14. 📊 Investors, P20 and S5.

• Prediction and band: an AI application company exits through insolvency or a documented wind-down (25%; range 15–40%). Segment A cover changing before claims remains pending the renewal sample. Severity: high. Probability: low.

• Exposure: reserve adequacy for retained AI loss at portfolio companies, measured in quarters of runway.

• Mitigating actions: diligence teams add endorsement-schedule, alert-ownership and indemnity questions to first-quarter 2027 portfolio reviews. Board observers request each company’s renewal schedule before its renewal date, using only information the company may share.

• Residual exposure: confidential settlements can hide retained loss from every review.

Most residuals trace to conduct fixed before the window opens: dated findings, signed caps and issued denials.

XXI. What to Watch: Dated Checkpoints Through March 31, 2027

The checkpoints below come from dockets, calendars and statutes with fixed dates. P6 is the dominant fork: a first complaint against an independent enterprise deployer activates S2 and S3, while silence by March 31, 2027 favors the account that liability stays with developers.

• November 24, 2026 (proposed): the sale hearing in the Character bankruptcy tests S1 and P4, with closing expected in late November or early December.

• December 1, 2026: the renewal sample window opens for P9 through P11, P14 and S5.

• January 1, 2027: the largest renewal cohort binds, and Colorado’s SB 26-189 takes effect with enforcement stayed pending rulemaking.

• January 3, 2027: the 119th Congress ends, and the AI LEAD Act needs reintroduction for P15.

• January 2027: most state legislative sessions open, which starts the P21 enactment window.

• January 31, 2027: P1 and P4 resolve.

• March 31, 2027: every remaining dated prediction resolves, and S3 runs 90 days from service of any P6 complaint.

The first deployer complaint and the first coverage filing decide which branch the next Snapshot starts from.

XXII. Conclusion: The Documented Response Decides Who Pays

Agentic common law starts from three questions the law has always asked: who knew, who controlled and who authorized. Agents scatter the answers, and the four doctrines that depend on them break in predictable directions. The breaks produce six competing duties, which play out across four stages on two clocks.

The 2026 incidents show the collisions starting on developers that deploy their own products. Carriers apply the structure at each renewal and then meet it again as plaintiffs and as AI users. State attorneys general apply it without waiting for a federal order, while lawmakers at both levels decide whether duties expand or preemption narrows them. The operational controller, the named defendant and the final loss-bearer can still be three different parties.

The decisive signal is P6 (45%; range 30–60%): a complaint naming an independent enterprise deployer by March 31, 2027. A filing confirms that loss is moving toward the enterprise deployer, where the four doctrines point, and no filing favors the rival account that liability stays with developers. The order in which each segment meets the two clocks then decides whether cover follows.

Working With MindCast

A public paper maps the doctrines, and a commissioned engagement maps the client’s own seat. MindCast offers a complimentary pilot before any retainer, built on public filings, policy forms and endorsements plus one deployment’s permission list.

The pilot delivers a harm-pathway map and an exclusion map. The pilot also places the client in the coverage and claim matrix and supplies Snapshot Predictions with odds, windows and falsifiers. A retained engagement under nondisclosure then models the client’s own regulators, plaintiffs, carriers and counterparties. Deliverables include an alert-ownership map, a retention and revocation review, an indemnity and recovery review and mitigation keyed to each checkpoint.

Carriers use the work to test whether an insured answered the six competing duties before binding. Coverage and subrogation counsel use it to see which doctrine governs a claim before the first ruling. Contact mcai@mindcast-ai.com. See predictive game theory simulations at our corporate site www.mindcast-ai-simulation.com.

Annotated MindCast Corpus

• FTC Investigation of Anthropic and OpenAI. Forecasts the federal theory that accelerates the pathways mapped here and names coverage conditions as a transmission channel.

• Agentic Foreseeability and the Modern Duty of Care for AI Agents. Supplies the four-question assessment and the foreseeability doctrine the four agentic doctrines extend.

• Who Writes AI Insurance Coverage Conditions. Supplies the exclusion filings and coverage conditions.

• Who Absorbs Correlated AI Loss. Supplies the aggregation and number-of-occurrences stakes.

Sources

Cases and proceedings

• MacPherson v. Buick Motor Co., 217 N.Y. 382 (1916).

• Palsgraf v. Long Island Railroad Co., 248 N.Y. 339 (1928), including the dissent of Judge Andrews.

• H.R. Moch Co. v. Rensselaer Water Co., 247 N.Y. 160 (1928).

• United States v. Carroll Towing Co., 159 F.2d 169 (2d Cir. 1947).

• Overseas Tankship (U.K.) Ltd. v. Morts Dock & Engineering Co. (The Wagon Mound No. 1), [1961] A.C. 388 (P.C.).

• Escola v. Coca-Cola Bottling Co., 24 Cal. 2d 453 (1944) (Traynor, J., concurring).

• The T.J. Hooper, 60 F.2d 737 (2d Cir. 1932).

• Moffatt v. Air Canada, 2024 BCCRT 149.

• United States v. Bank of New England, 821 F.2d 844 (1st Cir. 1987), cert. denied, 484 U.S. 943 (1987).

• Sall ex rel. Sall v. T’s, Inc., 281 Kan. 1355, 136 P.3d 471 (2006).

• Ybarra v. Spangard, 25 Cal. 2d 486 (1944).

• Raine v. OpenAI, Inc., No. CGC-25-628528 (Cal. Super. Ct. San Francisco, filed August 26, 2025).

• Amazon.com Services, LLC v. Perplexity AI, Inc., No. 26-1444 (9th Cir. August 4, 2026).

• In re Character Technologies, Inc., No. 26-11603 (Bankr. D. Del., filed October 12, 2026), with the U.S. Trustee case document and “Just 5 Years In, Interactive Chatbot Platform Character.AI Files for Bankruptcy” (TechCrunch, October 13, 2026), and the proposed sale schedule in “Disney Has Not Bought Character.AI as a Whole” (Neoteo, October 2026).

• Character.AI and Google agreement to resolve five family suits, reported in “Character.AI, Google Agree to Settle Teen Chatbot Harm Lawsuits” (Bloomberg Law, January 8, 2026).

• Tumbler Ridge suits against OpenAI (N.D. Cal., 7 filed April 29, 2026 and 30 more in September 2026), described at Edelson PC, AI Lawsuits.

• Nippon Life Insurance Company of America suit against OpenAI (March 2026), reported in “Agentic AI Liability Fuels Issues Reaching Beyond the Law’s Edge” (Bloomberg Law, July 8, 2026).

• Businessowners insurer declaratory action over AI-handled phone orders, reported in “Insurer Disputes Coverage for BIPA Suits Over Food Phone Orders Handled by AI” (Mealey’s), covering the Illinois filing.

• Estate of Lokken v. UnitedHealth Group, Inc., No. 23-cv-3514 (D. Minn.), discovery order of March 2026, discussed in “Court Allows Discovery Into Insurer’s Use of AI to Deny Claims” (Hunton Andrews Kurth).

• Ace American Insurance Co. v. Congruity 360 and Trustwave Holdings (2025), reported in “Cyber Insurer Sues Policyholder’s Cyber Pros” (National Law Review, October 13, 2025).

• Subrogation ruling for a vehicle-valuation software maker, reported in “Software Co. Freed From Insurer’s $2.5M Indemnity Dispute” (Law360, July 22, 2026).

• xAI v. Colorado (D. Colo., filed April 9, 2026), with the Department of Justice intervention described in “DOJ Joins xAI in Lawsuit Challenging Colorado AI Act” (Jenner & Block, April 29, 2026).

• Mobley v. Workday, Inc. (N.D. Cal.), discussed in the Bloomberg Law article cited above.

Federal action and advocacy

• AI LEAD Act, S. 2937 (introduced September 29, 2025), Senate press release of Senator Dick Durbin, with the section-by-section summary of sections 201 and 302.

• Executive Order 14365, “Ensuring a National Policy Framework for Artificial Intelligence” (December 11, 2025), sections 3 and 8(b), and the Department of Justice AI Litigation Task Force of January 9, 2026, discussed in “Inside the DOJ’s New AI Litigation Task Force” (Baker Botts).

• “Compromise Necessary but Not Sufficient for AI Preemption” (Washington Legal Foundation, June 22, 2026).

• Consumer Federation of America, “Complaint and Request for Investigation” on character-based therapy chatbots(June 2025).

• Federal Trade Commission section 6(b) orders on companion chatbots (September 11, 2025), discussed in “AI Companion Bots: Top Points From Recent FTC and Government Actions” (DLA Piper, September 15, 2025).

• Simpson Thacher, “Policyholders Developing AI-Based Bad Faith Claims” (August 28, 2026).

State enforcement

• Kentucky Attorney General suit against Character Technologies, Inc. (January 8, 2026).

• Pennsylvania State Board of Medicine suit against Character Technologies, Inc. (May 1, 2026).

• Florida Attorney General suit against OpenAI and its chief executive (June 1, 2026).

• Iowa Attorney General, announcement of the OpenAI suit (October 14, 2026), naming the coordinated state actions of that day, with KCTV’s report of Kansas joining seven other states.

• “Attorney General Ken Paxton Sues OpenAI and CEO Sam Altman” (Texas Attorney General, October 23, 2026).

Books and articles

• Oliver Wendell Holmes Jr., The Common Law (1881).

• Oliver Wendell Holmes Jr., “The Path of the Law,” 10 Harvard Law Review 457 (1897).

Restatements, statutes and rules

• Restatement (Second) of Torts sections 323 and 324A (American Law Institute, 1965).

• Restatement (Third) of Agency sections 1.04 comment e, 2.03, 3.03 and 3.11 (American Law Institute, 2006).

• Restatement (Third) of Torts: Products Liability sections 2, 10 and 11 (American Law Institute, 1998).

• California Civil Code section 1714.46.

• California Senate Bill 243, companion chatbots (Chapter 677, Statutes of 2025; effective January 1, 2026).

• Colorado Senate Bill 26-189 (signed May 14, 2026; effective January 1, 2027), repealing and replacing Senate Bill 24-205, as summarized by Buchalter.

• Federal Rule of Civil Procedure 37(e).

• 11 U.S.C. section 362, including the police-power exception in section 362(b)(4).

• Electronic Signatures in Global and National Commerce Act, Pub. L. 106-229 (2000), 15 U.S.C. section 7001(h) on electronic agents.

• Uniform Electronic Transactions Act section 14, automated transactions (Uniform Law Commission, 1999).

Incident, regulatory and market sources

• AI Incident Database, “Incident Report 2026: May, June, July”, covering Incidents 1471 through 1618.

• National Association of Insurance Commissioners, 2026 Report on the Cybersecurity Insurance Market (2025 reporting year).

• “Discover the Endorsement and Exclusions Filing Trends for 2026” (Independent Insurance Agents & Brokers of America, 2026).

Companion works: Regional Clean Compute Anchors (Installment One) · What Quantum Computing Contributes to Clean Compute Clusters (Installment Two) · How Quantum Computing Can Improve AI (Installment Three)

War game editions: Clean Compute Anchor War Game Executive Deck · Interactive Clean Compute Anchor War Game

Executive Summary

Core insight. A clean energy host’s move does the most while the instrument that will govern it is still unwritten. Once a district sets terms by negotiation, a county decides the first project or a host signs a tenancy, the same move can shape the next deal but cannot change the first one.

Thesis. The board carries five exposures. Two of them lost a route before any regional anchor filed: unconditioned allocation at Grant County in April 2025 and the refurbishment tax route on July 1, 2026. Four exposures still turn on decisions dated 2027 to 2029, and those two are among them.

The 105-day legislative session convenes January 11, 2027.

Grant County’s 2027 relief line, the first county decision and the first federal co-location order follow. The quantum tenancy has no closing date and stays open until a host signs.

What distinguishes MindCast. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through MindCast AI Proprietary Cognitive Digital Twin Foresight Simulations (MP CDT FS). Behavioral Economics supplies the decision rules. Game Theory supplies the payoff structure. Predictive simulations emerge from the combination.

The war game holds the three installments’ Cognitive Digital Twins (CDTs) fixed and changes one host’s move at a time.

What the paper does. The war game replays six dated turns, maps how the game evolves through three successor states after the 2027 session, plays five alternative moves against the same board and isolates the clock still most open to conduct. The Interactive Clean Compute Anchor War Game lets the reader take one of five seats: hydro host or reactor host, quantum host, legislator or lender. The Clean Compute Anchor War Game Executive Deck carries the board, the turns and the successor states in seventeen slides.

The war game carries the seventeen Simulation Predictions from the three installments and adds no new bands.

• 🏛️ Policymakers. The session’s four definitions decide whether hydro anchors negotiate under stated criteria or case by case; a nuclear-only bill leaves the likeliest first agreement outside the statute.

• 💼 Executives. Written import confirmation before an announcement changes what the county reads; after the announcement, the same request rides on deposits already paid.

• ⚖️ Counsel. A co-location reopener and a label-ownership clause cost little before execution and nothing can add them after.

• 📊 Investors. Read an anchor as five exposures. Allocation and tax have each lost one route. Four exposures turn on forks inside the next thirty months, and the quantum tenancy has no date.

Takeaway. The board shows which moves still secure power and terms, which ones only document what a host could have done, and on what date the line falls.

How the War Game Works

A paper narrates moves someone else made. A war game hands over the pieces: the reader takes a seat, makes a move and watches the other actors respond. The Interactive Clean Compute Anchor War Game runs every move in this paper, and the executive deck shows the same board in summary.

The war game runs in five steps. It sets the board with the Cognitive Digital Twins from the three installments, from districts and counties to vendors and lenders. It dates the clocks by naming each exposure and the public event that closes it.

It replays the turns and maps how each fork evolves the game. It then plays alternatives one move at a time and drops injects onto the board to read which Simulation Predictions move.

Behavioral Economics supplies the decision rules. Counties react to the salient label, so an announcement draws review that a measured filing at an energy site does not. Districts short of transmission prefer negotiation to a rule, and hosts overbuild quantum islands because the label is salient to lenders while utilization is not.

Game Theory supplies the payoffs. A host gains from a rule it can file against, and a district gains from discretion until relief lines energize. A vendor earns on each placed island while a host earns on utilization, and predictive behavior emerges from the combination.

Takeaway. The difference between the replay and the alternative is the finding, shown instead of argued.

I. A Host’s Move Buys the Most Before the Governing Instrument Is Written

A clean energy host faces five decisions made by five different authorities, and each authority writes its terms once. The district allocates import capacity and the county classifies the use. The legislature fixes the tax route and the definitions, FERC assigns co-location cost and the tenancy assigns quantum output. A host’s move changes the terms only while the authority that writes them has not yet written.

The mechanism runs through information. A district that receives a written import request with no announcement attached studies a load; a district that reads an announced operator agreement studies a precedent. A county that holds a burden exhibit before its first decision has a measured test to apply. A legislator who sees a live hydro project before drafting writes a definition that reaches it.

After the instrument is written, the same move changes less. A burden exhibit filed after a county pause becomes evidence for an appeal rather than the basis of the first reading. A label-ownership clause negotiated after the first label exists becomes a dispute rather than a term.

Two exposures depend on conduct and three on documents. Allocation, accessory use and quantum tenancy respond to what a host files and signs. Tax classification and co-location respond to what a statute and an order say, and a host’s move there is to put its facts in front of the drafter before the text is fixed.

Takeaway. Timing is the control. The move does not change between the early and late versions; the authority’s position does.

II. Clean Energy Hosts Carry Five Exposures, and Each Closes on Its Own Date

Each exposure has two dates: when it originates and when a preventive move stops being available.

Allocation. Does the host hold firm import capacity at anchor load? The exposure originates with a host’s first import request at an existing interconnection. Grant County’s public utility district capped data center load growth in April 2025 on transmission grounds, which closed the window for unconditioned allocation there.

Relief lines are due in 2027 and 2029. A statewide definition of added load would make the remaining allocation criteria predictable. The definition cannot create transmission headroom, so allocation stays subject to available capacity and the district’s service authority.

Accessory use. Does the county read the anchor as an accessory use at an energy facility or as a new data center? The exposure originates with the first local filing. The preventive window closes when the first county decides a modular or quantum-equipped project, which Installment One places inside 2029.

Texas showed on September 21, 2026 that a state executive can close the window statewide.

Tax classification. Does compute added inside an existing facility qualify for the data center exemption? Senate Bill 6231ended the refurbishment route on July 1, 2026 and kept the new-construction route. Whether an anchor falls under either route is a classification question the 2027 session can answer.

Co-location. Who pays for transmission a reactor no longer uses for export when compute sits behind the meter? The exposure originates with any compute agreement at a nuclear plant. FERC’s October 9, 2026 order left retail curtailment with the states, and the window closes when a co-location order assigns the cost.

Quantum tenancy. Who owns the labels and the models trained on them, and does the island install before a tenant exists? The exposure originates with the first island quote. No instrument dates its close; a host closes it by signing.

Two of the five have already lost a route for Washington hosts: unconditioned allocation at Grant and the refurbishment tax route. Both still face a later decision, as do accessory use and co-location, so four exposures turn on dated forks. The quantum tenancy stays open-ended.

Takeaway. The board holds five exposures. Two have lost a route, four turn on forks inside the next thirty months, and the quantum tenancy has no date.

III. Six Turns From the Reactor Announcement to the 2027 Forks

The replay steps through what hosts and authorities actually did and how each clock moved. Six turns run from October 2024 to the forks that resolve between 2027 and 2029.

• Turn 1, October 2024 to April 2025: the reactor route opens and a district caps load.

• Turn 2, April to July 2026: the tax route narrows and modules break ground.

• Turn 3, September 2026: a state halt and the first quantum computers inside production data centers.

• Turn 4, October 2026: DARPA dates the research tenant and FERC moves the forum.

• Turn 5, November 2026 to April 2027: the session decides four instruments.

• Turn 6, 2027 to 2029: relief lines, first agreements and the hub-or-island race.

Turn 1 · The reactor route opens and a district caps load. In October 2024 Amazon invested in Energy Northwest’s planned Xe-100 project near Richland, with construction planned by 2030. In February 2025 Duke University’s Nicholas Institute estimated that 76 gigawatts of new load could join the grid at 0.25 percent curtailment. In April 2025 Grant County capped data center load on transmission grounds.

Clocks after the turn: allocation closed for unconditioned requests at Grant; the other four open.

Turn 2 · The tax route narrows and modules break ground. In April 2026 a Google Quantum AI and Caltech team showed in simulation that quantum oracle sketching saves memory on classical data while data loading still dominates runtime. On July 1, 2026 the refurbishment route to the data center exemption ended. On July 27 Energy Vault broke ground on an 8-megawatt modular campus at Snyder, Texas.

Clocks after the turn: tax closed on the refurbishment route; accessory use open with modules still unclassified.

Turn 3 · A state halt and the first quantum computers in production data centers. On September 1 Equinix and Diraq announced a planned quantum computer drawing under 20 kilowatts inside a Sydney data center. On September 21 Texas halted data center environmental permits. On September 23 IonQ announced a 256-qubit system coupled to NVIDIA hardware for installation in 2027.

Clocks after the turn: accessory use reducible at rising cost now that a state veto is on the board; quantum tenancy reducible as island quotes arrive.

Turn 4 · DARPA dates the research tenant and FERC moves the forum. On October 7 DARPA moved four more teams into the final stage of its Quantum Benchmarking Initiative, which asks whether any approach reaches utility scale by 2033. On October 9 FERC rejected PJM’s curtail-first plan for large loads and left retail curtailment with the states.

Clocks after the turn: co-location reducible with the docket open and the state forum named.

Turn 5 · The session decides four instruments. The Washington Legislature convenes its 105-day session on January 11, 2027. Four instruments are available: a definition of added load at existing interconnections, a definition of accessory computing use keyed to burden, a tax classification fix and a functional protected-load test.

Clocks after the turn: allocation, accessory use, tax and the protected-load half of quantum tenancy turn on the session.

Turn 6 · Relief lines, first agreements and the hub-or-island race. Grant County’s first relief line is due in 2027 and the second in 2029. Installment One’s first anchor agreement resolves by December 31, 2028. Installment Two’s hub-or-island race resolves by December 31, 2031.

Clocks after the turn: four forks decided; quantum tenancy open until a host signs.

Takeaway. The turns that closed windows were a district cap and a tax statute, not a county decision; the turns ahead are a session, a relief line and an order.

IV. How the Clean Compute Anchor Game Evolves Through 2031

A forecast holds only inside the game that produced it. The three installments are Foresight Simulations, and each fixed its Simulation Predictions after most of the rules had formed. The war game carries those predictions forward into a game that keeps changing. Readers therefore need to know how the game changes and when a change requires a prediction to be revalidated.

Evolution and Mutation Name Two Different Kinds of Change

MindCast calls an anticipated change in the game evolution. A scheduled session, a dated relief line or a pending federal order changes the game in a direction the simulation already maps. The change can reach the players and their strategies, the payoffs and rules, or the timing. The simulation forecasts evolution, so the Simulation Predictions carry through it with a status change and no new band.

Mutation is reserved for the exception. A mutation is an anomalous or temporary departure the mapped paths do not cover, such as a buyer outside the modeled set or a pause that lifts before any term is written. A mutation suspends the Simulation Predictions it touches until the simulation revalidates them against the new state.

Most events change neither. An event that adds information without changing anyone’s moves is a continuation, and the game stays as it was. The April 2026 oracle-sketching result showed where quantum saves memory but left every seat’s strategies intact.

Takeaway. The game evolves on dated forks the simulation maps; mutation names the anomaly, and continuation names the event that changes only what players know.

The Game Has Evolved Four Times Since 2024

Grant County’s April 2025 cap was the first evolution. Allocation changed from a request a district answers to a negotiation a district controls, and the host lost the move of an unconditioned request.

Senate Bill 6231 was the second. The July 1, 2026 statute removed the refurbishment route, so a host lost one tax strategy and kept the new-construction route.

The September 2026 quantum announcements were the third. Equinix and Diraq brought a quantum processor into a production data center, and IonQ dated a coupled system for 2027. A new seat joined the board: a vendor that earns on each placed island.

FERC’s October 9, 2026 order was the fourth. The order changed the forum rather than the outcome, moving curtailment and continuity terms into state proceedings. DARPA’s October 7 final-stage selections changed timing in the same month by dating the research tenant’s budget question to 2033.

Texas’s September 2026 permit halt was a continuation for Washington hosts. The halt changed no Washington rule, but it showed counties and lenders that a state executive can close the accessory-use window statewide.

Takeaway. Two of the four evolutions closed windows, and the other two added a seat and moved a forum. None changed the payoff a host earns from moving first.

Each Fork Carries Five Entries

Every fork on the board is mapped with the same five entries before any Simulation Prediction attaches to it:

1. What the event changes in the game: the players and strategies, the payoffs and rules, or the information and timing.

2. Which paths it opens and which it closes.

3. Which seat positions survive, convert or strand.

4. Which seat gains or loses a move.

5. Which Simulation Predictions apply in the resulting state, and with what status.

The order matters. The simulation fixes the successor state first and attaches Simulation Predictions second, so two branches never rely on conflicting assumptions about the same actor’s authority.

Takeaway. A fork without its five entries gives a reader odds on the event and no map of what follows it.

The 2027 Session Produces Three Successor States

The session that convenes January 11, 2027 is the dominant fork, and each outcome produces a different game.

Burden-keyed package passes: the statewide criteria route. The rules change. Allocation criteria at existing interconnections become predictable, and counties apply a measured burden test. The statute creates no headroom, so each allocation stays subject to available capacity and the district’s service authority.

Hydro hosts’ import requests and burden exhibits convert from evidence into filings against a stated standard. Hosts gain the move of filing against published criteria, and districts lose discretion over the terms of the first agreement while keeping authority over capacity. One S-4 (34–46%) resolves true, and One P-1 (45–57%) moves inside its window under stated criteria.

Nuclear-only definition passes: the split route. The rules change at one source only. Reactor hosts gain a statutory template that waits for operation in the mid-2030s.

Hydro hosts stay in negotiation with their district, so their positions survive but do not convert. Districts keep their discretion over the likeliest first agreement.

Two outcomes separate on this route. One S-4 resolves true if the nuclear-site definition treats computing equipment by measured burden, because the forecast reaches any existing generation facility. A statewide anchor framework still does not form, and One P-1 runs by negotiation at hydro sites. The simulation places the split route as the likeliest 2027 outcome.

No definition passes: the negotiated route. The game continues on its current rules. The first district negotiation becomes the template for every later anchor, and a host that announced before confirming import capacity is the most likely to be stranded.

Districts gain the move of setting the template, and hosts keep only the move of negotiating first. One S-4 resolves false.

Takeaway. The session decides which game the rest of the board plays in, so a host’s filings before it carry into whichever successor state arrives.

The Board Keeps Only the Distinct States

The session, the relief line and the co-location order produce eighteen combinations on paper. The board keeps three successor states, because the relief line and the co-location order change timing and cost inside each route rather than changing the route.

A relief line that energizes on schedule shortens the wait in every state. A slip moves allocation at Grant to the 2029 line in every state. A FERC order that assigns transmission cost to co-located load triggers the reopener in reactor agreements whichever route the session chose. A playable board maps the reachable states that differ in strategy, not every arithmetic combination.

Takeaway. Three successor states carry the board through 2031, and the remaining forks modify timing and cost within them.

Two Injects Are Mutations, and Four Are Evolution

The Interactive Clean Compute Anchor War Game’s six injects test the board’s response to shocks. Four are evolution of a known type. A Bonneville statement on added load makes allocation criteria more predictable, and a reactor slip past 2034 changes timing. A materials model trained on quantum-computed labels opens the labels channel, and a processor with no cryogenic plant makes local ownership ordinary.

Two are mutations. A county pause on modules at an operating energy site is temporary: label review stalls the accessory-use clock without writing a term, and the game resumes when the pause lifts. A hyperscaler that acquires regional anchor sites below 50 megawatts is anomalous: the customer tier the series models collapses along with the shared-hub route. Under either mutation, the affected Simulation Predictions are suspended until the simulation revalidates them.

Takeaway. Mutation is the exception the board flags, not the default label for change.

What Game Evolution Means for Each Audience

🏛️ Policymakers. The session chooses among three games, not three bills. Only the burden-keyed package gives hydro anchors stated criteria where the first agreement is likeliest. A nuclear-only definition can resolve One S-4 and still leave hydro anchors outside the framework.

💼 Executives. File before the session so the filings convert in the statewide criteria state and survive in the split state. An announcement made before confirmation is the position most likely to be stranded on the negotiated route.

⚖️ Counsel. Draft agreements to read correctly in all three successor states. A co-location reopener and a label-ownership clause carry across every state because they respond to forks that evolve inside each route.

📊 Investors. Underwrite a Washington anchor by successor state rather than by headline. Treat a county pause or a hyperscaler roll-up as a mutation that suspends the affected Simulation Predictions, not as evidence against the route.

Takeaway. The Simulation Predictions hold while the game evolves along mapped paths, and the board shows which state each one belongs to.

V. Five Alternative Moves and How the Board Responds

Each alternative changes one move at one turn and runs the same actors against it. Alternatives carry no band and do not change any Simulation Prediction’s band.

1. Request written import confirmation before announcing.

Turn 3, hydro host. In the simulation, the district studies a load rather than a precedent, and the county later reads a filing with import rights in hand. Changes allocation and accessory use. Cost tier: low, legal and engineering review.

2. File a burden exhibit with the first local filing.

Turn 3, any host. Water, noise and generator hours sit in front of the county before any pause. Changes accessory use. Cost tier: low, measurement and filing.

3. Add a co-location reopener to every reactor-site agreement.

Turn 4, reactor host. A later order resets terms by contract instead of landing on one party. Changes co-location. Cost tier: low, transaction counsel.

4. Negotiate hub access before any island quote and write label ownership into the tenancy.

Turn 3, quantum host. The host keeps its reserved circuit and assigns the output before it exists. Changes quantum tenancy. Cost tier: low to medium, negotiation and counsel.

5. Draft the four instruments as one burden-keyed package.

Turn 4, legislator. Allocation, accessory use and tax classification move together and reach hydro anchors as well as the reactor site. Changes three clocks. Cost tier: medium, session time and coalition.

Two sequencing errors deepen exposure when the simulation runs them against the same actors. Neither describes a documented Washington transaction. An operator agreement announced before import confirmation draws county review as a data center and puts deposits ahead of allocation. An island accepted before a tenant exists carries $3.5 million to $4.5 million a year of fixed cost at 10 to 20 percent utilization.

Takeaway. Played on time, each move changes the evidence and the options at a stated cost; played late, it mostly documents what the host could have done.

VI. Session Filings Before the 2027 Session Are the Exposure Hosts Can Most Readily Move

The session clock is the one most open to conduct now. A working group drafts against the projects in front of it, and Behavioral Economics predicts the committee responds to the salient project, which in Washington is the reactor. A hydro host that files an import request and a burden exhibit before drafting begins puts a second project in front of the committee.

The simulation places a nuclear-only definition as the likeliest 2027 outcome, because the Richland development interest and the backlash against data center proposals near Cle Elum pull the working group in opposite directions. A nuclear-only definition leaves hydro anchors in case-by-case negotiation at the site where the first agreement is likeliest. A burden-keyed definition across generation types reaches all three anchor sources and costs the reactor site nothing.

Installment One carries the band. Washington adopts a definition treating computing equipment at an existing generation facility by measured burden rather than by the data center label by December 31, 2027 at 34–46%. The first anchor agreement at a Washington hydro or nuclear interconnection by December 31, 2028 carries 45–57%.

Under a burden-keyed definition the first agreement runs on stated criteria within available capacity. Without one it runs by negotiation, and the district’s first negotiation sets the terms for every later anchor.

• ⚖️ Counsel to hosts. Prepare the tax classification analysis for an existing-facility anchor before any bill text circulates, since the remedy differs by which route the anchor falls outside.

Takeaway. The main conduct move still open costs a host a filing now and decides whether the first agreement in Washington runs on stated criteria or by negotiation.

VII. What the War Game Means for Each Seat on the Board

Each seat on the board faces the same first question in its own vocabulary: which authority writes the next term, and has it written yet.

💼 Hydro and geothermal hosts. Allocation is the binding constraint whichever authority acts first. Request written import confirmation before any announcement, file the burden exhibit with the first local filing, and qualify an alternative geothermal or out-of-state hydro site so the district weighs losing the anchor.

💼 Reactor hosts. Federal co-location binds before any state question at a nuclear plant. Add a co-location reopener to every agreement before execution, and firm the first phase on hydro or grid service so capital stages to the construction permit rather than waiting for operation in the mid-2030s.

💼 Quantum hosts and vendors. An island installs in two quarters and a hub needs about six to serve three sites, so an island wins the race whenever a vendor offers placement terms. Negotiate hub access first, keep the reserved circuit, and assign label ownership in the tenancy before any label exists.

🏛️ Lawmakers. The four instruments pass separately or together, and only the package reaches hydro anchors. Draft them as one burden-keyed package and tie the protected-load test to function and recovery time rather than to the quantum label.

📊 Lenders. Allocation and permitting resolve on different dates. Stage funding to the executed allocation agreement at hydro sites and to the construction permit at reactor sites, and underwrite quantum at 10 to 20 percent utilization until a hub serves three sites.

Takeaway. Every seat’s best move follows from dating the authority that writes its next term.

VIII. Competing Hypotheses

Three accounts compete before any fork resolves.

Hypothesis A · Early filings set the terms. Hosts that file import requests and burden exhibits before the session put hydro projects in front of the drafters, and the definition reaches them. Falsifier: a burden-keyed definition passes with no hydro host having filed before drafting.

Hypothesis B · The salient project sets the terms. The working group drafts for the reactor regardless of filings, and hydro anchors negotiate case by case. Falsifier: a statewide definition across generation types passes in 2027.

Hypothesis C · Districts set the terms before the legislature. A district’s first negotiated allocation becomes the template before any definition takes effect. Falsifier: no district executes an anchor agreement before the session’s definitions take effect.

The war game favors Hypothesis B as the base case and shows Hypothesis A as the move that shifts it. Hypothesis C governs wherever the session passes nothing.

Takeaway. The hypotheses differ on who writes first, and each carries a dated falsifier.

IX. MindCast Simulation Predictions

The war game carries seventeen Simulation Predictions from the three installments and adds no new bands. P marks a Primary Simulation Prediction and S a Secondary Simulation Prediction. Seven lead the board.

• One P-1 · First anchor agreement at a Washington hydro or nuclear interconnection, first phase under 20 megawatts (45–57%). Window: December 31, 2028. Clock: allocation. Falsifier: no such agreement, or the first one sits above 20 megawatts.

• One P-2 · First county decision in the six regions lands on a firm-clean anchor under 20 megawatts (42–54%). Window: December 31, 2029. Clock: accessory use. Falsifier: the first decision concerns a campus or a project above 20 megawatts.

• One S-4 · Washington adopts a burden-based definition for computing equipment at generation sites (34–46%). Window: December 31, 2027. Clock: accessory use and allocation. Falsifier: no such definition in statute, rule or ordinance.

• Two P-1 · Shared hub serves three sites before a second island installs, given a multi-site research buyer (48–60% given the buyer). Window: December 31, 2031. Clock: quantum tenancy. Falsifier: a second clean-hosted island installs first.

• Two P-2 · The first clean-hosted island reports a model trained on its own labels before any energy metric (44–56%). Window: December 31, 2032. Clock: quantum tenancy. Falsifier: the first disclosure reports an energy or capacity metric.

• Three P-1 · The first qualifying quantum-improved AI result after October 28, 2026 is a labels result (62–74%). Window: December 31, 2031. Clock: quantum tenancy. Falsifier: the first qualifying result is a sampling or kernel result.

• Three S-4 · The first commercial AI service on quantum-produced data runs on shared or remote access (64–76%). Window: December 31, 2033. Clock: quantum tenancy. Falsifier: the first service runs on an operator-owned processor.

The remaining ten carry the bands printed in the installments.

• Installment One: S-1 60–72% · S-2 36–48% · S-3 40–52%

• Installment Two: S-1 78–88% · S-2 56–68% · S-3 66–78%

• Installment Three: P-2 70–82% · S-1 55–67% · S-2 64–76% · S-3 40–52% given authority

Takeaway. Moves, forks and injects change a Simulation Prediction’s status in the war game, never its band.

X. Risk Mitigation Layer

The actions below are analytic options for each audience and not legal, investment or fiduciary advice to any party. Severity and probability run on separate axes.

💼 Hydro host · One P-1 (45–57%).

• Exposure: an operator agreement signed before import confirmation carries equipment deposits and quarters of lead time against an allocation that may not arrive. Severity: high. Probability: moderate.

• Actions: the development lead requests written import confirmation at module load before any announcement, by the next district board cycle; the commercial lead qualifies an alternative site by mid-2027.

• Residual: a district can revise allocation rules after the agreement, and a later phase can trigger restudy.

🏛️ Lawmakers · One S-4 (34–46%).

• Exposure: a session that passes no definition, or a nuclear-only one, leaves the first hydro agreement to district negotiation and hands the legislature a precedent it did not write. Severity: moderate. Probability: moderate.

• Actions: legislative staff draft the four instruments as one burden-keyed package before the session’s policy cutoff; committee staff request burden data from pending hosts during the interim.

• Residual: Bonneville terms sit under federal authority and can override a state definition.

⚖️ Counsel at reactor hosts · One S-3 (40–52%).

• Exposure: a reactor-site agreement silent on co-location cost inherits whatever a later order assigns. Severity: high. Probability: moderate.

• Actions: transaction counsel adds a co-location reopener before execution; regulatory counsel documents the client’s assessment of the pending docket.

• Residual: FERC can adopt terms neither party proposed.

💼 Quantum host · Two P-1 (48–60% given the buyer).

• Exposure: an island accepted before joining a hub carries $3.5 million to $4.5 million a year of fixed cost at 10 to 20 percent utilization. Severity: high. Probability: moderate.

• Actions: the commercial lead negotiates hub access before accepting any island quote, by the first 2027 hosting cycle; the engineering lead keeps the reserved circuit sized for a later island.

• Residual: a vendor’s placement offer can make an island rational for one host and irrational for the region.

📊 Lenders · Two P-1 (48–60% given the buyer).

• Exposure: an underwriting case that credits an island with hub-level utilization overstates annual margin by the island’s fixed cost. Severity: moderate. Probability: moderate.

• Actions: the credit committee underwrites quantum at 10 to 20 percent utilization until a hub contract with three remote sites exists; the diligence lead requires the hosting agreement’s access terms before closing.

• Residual: utilization at the first hub may still fall short.

⚖️ Counsel to hosts and laboratories · Three S-4 (64–76%).

• Exposure: a tenancy silent on label ownership assigns the channel’s value by default once buyers take delivery remotely. Severity: moderate. Probability: moderate.

• Actions: transaction counsel allocates label ownership, derived-model ownership and publication rights in every access agreement now.

• Residual: a federal procurement clause can override contractual allocation for federally funded labels.

Takeaway. Each entry names a move that costs least before its clock’s authority writes and still costs something after.

XI. What to Watch

The dominant fork is the 105-day session that convenes January 11, 2027.

• January 11, 2027. The session convenes; bill text on added load, accessory use, tax classification and protected load moves One S-4 and One P-1.

• 2027. Grant County’s first relief line energizes or slips; headroom under the cap moves One P-1 and One S-2.

• Each FERC co-location order. Cost assignment at nuclear plants moves One S-3 and the reactor route.

• December 31, 2027. One S-4 resolves against statutes, rules and ordinances.

• December 31, 2028. One P-1 and One S-2 resolve against district minutes and operator releases.

• The first clean-hosted quantum installation. Hub or island decides Two P-1, given a multi-site buyer.

Takeaway. The session week is the dominant fork, and a host’s filings before it decide whether the hydro anchor is in front of the drafters.

XII. Conclusion

The war game’s dating discipline returns one answer at every seat: move before the authority that writes your next term has written it. A district cap and a tax statute closed two routes before any regional anchor filed. No host could have reopened either one afterward.

The counterfactual for hosts is the same import request filed in a different order. Filed before an announcement, it lets a district study a load and a county read a measured filing. Filed after, it rides on deposits and a salient headline. The session that convenes January 11, 2027 is the next authority to write, and the filings a host makes before it are the move the board shows is still open.

Working With MindCast

A commissioned war game dates the client’s own clocks, then plays the client’s moves against the same Cognitive Digital Twins. A hydro host learns which district and county terms its filings can still shape. A reactor host learns how its agreements read under each co-location outcome. A quantum host learns whether its region forms a hub or islands, and a lender learns which gates its capital should wait for.

Play the war game: the Interactive Clean Compute Anchor War Game runs Historical Replay, Counterfactual Play and Future Evolution. Readers take a seat, decide the session and the 2027 to 2031 forks and drop up to six injects onto the board. The Clean Compute Anchor War Game Executive Deck carries the board, the turns and the successor states in seventeen slides.

A complimentary pilot runs on public filings before any retainer. Contact mcai@mindcast-ai.com or visit www.mindcast-ai-simulation.com.

Appendix A. Selected MindCast Works

Clean Compute Anchor War Game Executive Deck. Carries the five-exposure board, the six turns, the three successor states and the five alternative moves in seventeen slides for a reader with ten minutes.

Interactive Clean Compute Anchor War Game. Runs the paper’s moves as play: the reader takes one of five seats, decides the 2027 to 2031 forks and watches the board and the Simulation Prediction statuses respond.

Regional Clean Compute Anchors, How Much Compute Each Clean Energy Source Yields and Where Washington Builds the First One (Installment One). Supplies the allocation, accessory-use, tax and co-location clocks and the four session instruments the board plays.

What Quantum Computing Contributes to Clean Compute Clusters, Tenant First, Training Data Second and Not the Processor Through 2032 (Installment Two). Supplies the hub-or-island race, the installation-speed asymmetry and the quantum host’s fixed cost.

How Quantum Computing Can Improve AI: Five Applications, Their Commercial Potential, and the Limits of Quantum Advantage (Installment Three). Supplies the label-ownership clause and the remote-delivery finding that separates product demand from island demand.

Regional Clean Compute Anchors, Quantum Tenancy and Quantum-Labeled Training Data: One Summary of the Three Clean Compute Anchor Installments. Supplies the seventeen Simulation Predictions the war game carries in one place.

AI Chip Export Enforcement War Game Shows When Verification Protects US Firms and When It Arrives Too Late. Supplies the war-game method this paper applies: dated clocks, replayed turns and alternatives that carry no band.

The Data Center Authorization Market: A 50-State Regulatory Atlas. Supplies the state evidence behind the accessory-use and allocation clocks.

The Model Data Center Authorization Code. Supplies the common language for the burden-keyed definition the legislator seat drafts.

Appendix B. Sources

• Energy Northwest and X-energy. Cascade Advanced Energy Facility, with Amazon investment announced October 2024. Turn 1.

• Duke University Nicholas Institute. Rethinking Load Growth. February 2025. Turn 1.

• Grant County Public Utility District. Data center load cap, April 2025, as reported by GovTech. Turns 1 and 6.

• Zhao and co-authors. Exponential Quantum Advantage in Processing Massive Classical Data. arXiv, April 2026. Turn 2.

• Washington State Legislature. Senate Bill 6231, 2026. Turn 2.

• Business Wire. Energy Vault Breaks Ground on Powered AI Infrastructure Campus in Snyder, Texas. July 27, 2026. Turn 2.

• Equinix. Diraq and Equinix to Deploy Australia’s First Quantum Computer Inside a Data Centre. September 1, 2026. Turn 3.

• Holland & Knight. Texas governor halts data center environmental permitting. September 2026. Turn 3.

• IonQ. IonQ to Advance Quantum Supercomputing by Bringing First QPU to NVIDIA Accelerated Quantum Research Center. September 23, 2026. Turn 3.

• DARPA. Four More Teams Enter Quantum Benchmarking Initiative Final Stage. October 7, 2026. Turn 4.

• FERC. Order of October 9, 2026 in Docket ER26-3515. Turn 4.

• Washington State Legislature. Session calendar. Turn 5.

Companion works: Regional Clean Compute Anchors · What Quantum Computing Contributes to Clean Compute Clusters · How Quantum Technology Changes the AI Data Center Energy, Compute and Environmental Calculus

Why This Installment Matters to Each Audience

Each audience will be asked, within a year, to fund or approve something described as quantum improving AI. The installment states which of five channels can deliver that improvement, in what order and for whom.

🏛️ Policymakers. Program officers and legislators will see quantum access proposals framed as AI acceleration. Only one channel produces AI output within the horizon, and it produces data rather than speed. A program that funds that channel with a data deliverable sets the procurement precedent the field lacks; a program that funds optimization or general AI spends a budget cycle on channels that stay inside measurement noise.

💼 Executives. Hosts, vendors and model builders face the same question in three vocabularies: what will a buyer pay for. The installment ranks the channels by date and value and states the decision rule, net quantum-AI value against the best classical alternative. Executives who sell processor placement on the strength of a quantum-AI product will find that product’s buyers taking delivery remotely.

⚖️ Counsel. The first commercial asset quantum produces for AI, on every route the simulation runs, is a dataset and the model trained on it rather than a machine. Access agreements written today assign that asset by silence. Counsel who allocate label ownership, derived-model ownership and publication rights before the first result exists decide who captures the channel; counsel who wait inherit a dispute.

📊 Investors. Every channel but one attracts a classical matching result within four quarters of publication. The installment supplies the survival test that separates a revenue line from a press release, and the one channel whose value moves to larger systems when classical methods catch up rather than disappearing.

Takeaway. The installment gives each audience the channel to fund, the asset to own and the test to apply, and Sections VII and VIII state what to conclude and what to do.

I. Quantum Improves AI Through Data First, and the Processor Channel Stays Closed

Quantum computing improves artificial intelligence by producing data that no classical method can produce, consumed by models that run on ordinary processors. Five channels carry that contribution. One of them pays within the horizon on the base route, two survive in narrow forms, one stays inside measurement noise and one works today without quantum hardware.

The first channel to produce a result a buyer will pay for is quantum-computed training labels for scientific models. The channel needs nothing undiscovered, it has a consumer already training on worse labels, and its value belongs to whoever owns the trained model. Every other channel either waits on a discovery or loses to a classical match within a year.

Thesis. Quantum improves AI first through labels for scientific models on strongly correlated matter. On the modeled base route the first qualifying publication arrives by 2028 and licensing follows by 2029, and P-1 carries the window through 2031 with a twelve-month survival test.

Sampling-based synthetic data survives in one physics-structured case and nowhere general. Kernels and pipeline optimization stay inside noise, and quantum-inspired software delivers today and gets mislabeled as hardware.

The actor that converts the labels channel into a market is a research buyer who procures labeled data as a deliverable, and that buyer takes delivery remotely.

Product demand therefore does not create processor demand. A buyer of quantum-produced labels needs the output and not the machine, so the first commercial quantum-AI service runs on shared or remote access. Installment Two’s hub-versus-island contest resolves on that fact, and Installment One’s reserved island circuit stays reserved until a tenant names a workload that needs local coupling.

A quantum capability becomes infrastructure demand only after four separate decisions: a validated gain, a buyer’s value test, a procurement that assigns ownership and a delivery mode. Classical catch-up feeds back to the first link and sorts the channels by whether their value survives a match.

Why now. On October 7, 2026 the Defense Advanced Research Projects Agency (DARPA) moved four more companies into the final stage of its Quantum Benchmarking Initiative, and the initiative asks whether any approach reaches utility scale by 2033. IBM’s sample-based quantum diagonalization already runs molecular ground-state problems at 50 to 100 qubits beside classical supercomputers. IBM’s roadmap reaches about 200 logical qubits by 2029.

A hybrid study, Hybrid Boson Sampling-Neural Network Architecture for Enhanced Classification, paired boson sampling with a neural classifier and reported gains over linear and sigmoid kernels on four datasets. A Google Quantum AI and Caltech paper of April 2026 titled Exponential Quantum Advantage in Processing Massive Classical Data showed in simulation that a method called quantum oracle sketching matches classical accuracy on sentiment and single-cell data with four to six orders of magnitude less memory. The method uses under 60 logical qubits. Neither result beat a tuned classical baseline on accuracy or runtime, and the gap between them and a buyer-paid result is the subject of this installment.

What distinguishes the MindCast approach. Vendors publish roadmaps and physicists publish benchmarks. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through MindCast AI Proprietary Cognitive Digital Twin Foresight Simulations (MP CDT FS). Each Cognitive Digital Twin (CDT) models how a named institution decides as evidence and rules change, and Section V names the twelve this installment runs.

Behavioral Economics supplies the decision rules: the hardware label is salient to hosts, lenders and counties while the software delivering the gain is not, and laboratories buy machine access until a replicated result teaches them the resale value of labels. Game Theory supplies the payoff structure: hyperscalers earn publication priority and vendors earn placement. Classical researchers earn by matching every general claim, and buyers pay only against the best classical alternative. Predictive simulations emerge from the combination.

What the paper does. Section II ranks the five channels by date and value against the anchor’s phases. Section III states why labels pay and who owns them. Section IV states where sampling and kernels survive.

Section V states the actor system, Section VI carries the MindCast Foresight Simulation Predictions, Section VII states the strategic insights for each audience and Section VIII carries the Risk Mitigation entries.

Simulation Prediction highlights. Six Simulation Predictions in Section VI.

• P-1. Among qualifying results published after October 28, 2026 the first model trained on quantum-produced data to beat the strongest classical baseline is a labels result for a materials or chemistry model. Window through December 31, 2031. 62–74%.

• P-2. No quantum-sampled synthetic dataset improves a named general AI benchmark over the best classical generator at equal compute cost, through December 31, 2030. 70–82%.

• S-4. The first commercial AI service built on quantum-produced scientific data runs on shared or remote quantum access and not on an operator-owned processor, through December 31, 2033. 64–76%.

Stakeholder readout.

• 🏛️ Policymakers. Write the first quantum access procurement for a chemistry or materials program with labeled data as the deliverable, before DARPA’s first final-stage report. Tie any state quantum incentive to a named application class and a stated classical baseline (S-2, S-3).

• 💼 Executives. Sell the anchor’s quantum role as hub access and research tenancy, not processor ownership. License the first published quantum-labeled model regardless of its host (P-1, S-1, S-4).

• ⚖️ Counsel. Allocate label ownership, derived-model ownership and publication rights in every quantum access agreement now, before any procurement precedent exists (S-3).

• 📊 Investors. Credit a channel’s revenue only after a replicated result survives twelve months without a classical match. Underwrite labels revenue per validated candidate and nothing per megawatt (P-1, P-2).

II. Five Channels Ranked by Date and Value

Quantum’s contribution to AI arrives through five channels, and the ranking turns on two questions: does the channel need anything undiscovered, and does a buyer already exist for its output. The section states each channel’s mechanism, its unit of improvement and its delivery mode, then places all five against the anchor’s phases.

The Inference Wall Sets the Boundary

Quantum processors do not run inference or training within the horizon. Loading classical data into a quantum state costs about as much as the computation it would accelerate, reading the answer out collapses the state to a few bits, and error correction multiplies every logical gate by hundreds of physical ones. Installment Two states the wall in full, and this installment takes it as the boundary: every channel below delivers something to a classical model, and none replaces the processor the model runs on.

The Channels

The labels channel opens in the anchor’s year-three phase and carries through the horizon. Kernels close before the island phase begins. Software runs the whole span on hardware the anchor already owns.

Why the Order Holds

K1 ranks first because it combines a demonstrated capability with an existing consumer. Sample-based diagonalization already computes ground-state energies at sizes exact classical methods cannot store, and materials foundation models at Google, Microsoft and Meta already train on classical density-functional labels. The channel’s only missing piece is a label set above a thousand points on a correlated system, which the base route places by 2028 and the alternative route two to three years later.

K2 ranks second because its separation is proven and its application is not. Boson sampling and random-circuit sampling show distributions classical samplers cannot reach, and no research group has identified a distribution that is both provably hard and useful as training data outside physics. The one plausible survivor is molecular conformer sampling, which is a K1 complement rather than a channel of its own.

K3 and K4 rank below because every general claim attracts a classical match. The dequantization line of work, beginning with Tang’s 2018 result on recommendation systems, matched each claimed exponential speedup with a new classical algorithm. Pipeline optimization faces a different wall: classical solvers already produce near-optimal plans in milliseconds, and the gap to optimal is one to three percent, inside the buyer’s measurement noise.

K5 ranks last by novelty and first by availability. Tensor networks and annealing-derived heuristics improve compression and scheduling on ordinary GPUs today. Most quantum productivity cited in corporate materials is this software, and MindCast’s companion Funded Solutions That Can Help Win Quantum-Coupled AI Data Centers Approvable already separates its value from hardware. The separation matters here because a host that credits K5 gains to a processor will build one it does not need.

Takeaway. One channel pays on the base route, one survives narrowly, two stay inside noise and one needs no hardware. The order follows from what each channel needs and who already buys it.

III. Labels Pay, and the Tenancy Decides Who Collects

The labels channel is the one that pays, and its value accrues to whoever owns the model trained on the labels. The section states the mechanism, the decision rule a buyer applies and the ownership question the first contracts leave open.

The Mechanism

Classical simulation of electronic structure is approximate once electrons correlate strongly. Catalysts, battery electrolytes, superconductors and reaction dynamics all live in that regime. A fault-tolerant quantum machine computes those energies to chemical accuracy at sizes classical methods cannot reach. The outputs become labels for neural network potentials and materials foundation models, and the model proposes candidates while the quantum machine validates the few worth the expense.

The value of a label is its cost to produce classically. Where coupled-cluster methods reach the same accuracy, the label is worth its classical cost and the quantum machine adds nothing. Where classical cost grows exponentially with system size, the label has no classical substitute and its value is the discovery it enables. The channel’s frontier therefore moves: when a classical method matches quantum accuracy on the first system, the value shifts to the next size up rather than disappearing.

The Decision Rule

A buyer applies one test. Net quantum-AI value equals the value of the improved AI output less the incremental cost of quantum delivery. The comparison runs against the best practical classical alternative and includes every delivery cost from data preparation through integration.

A positive result justifies commercial evaluation. A positive result does not by itself justify a local processor, because labels travel as data and the buyer takes delivery from wherever the machine sits.

Game Theory supplies the payoff: hyperscalers earn publication priority by publishing the first labels result, so the first qualifying paper carries a hyperscaler affiliation before any laboratory or clean-hosted island reports one. Behavioral Economics supplies the error: laboratories buy machine access with no data clause until a replicated result teaches them the resale value of labels, and only then do they procure data as a deliverable. Predictive behavior emerges from the combination, and P-1, S-1 and S-3 state it.

The Ownership Question

Access agreements in the current market assign output by silence. A laboratory that buys machine time under a standard access contract leaves the labels with whichever party holds them when the program ends, and the model trained on them with whichever party trained it. The first procurement that names labeled data as a deliverable will set the precedent, and every agreement signed before it will be read against that precedent.

Counsel therefore hold the distribution decision for the channel. Label ownership, derived-model ownership, publication rights and export-control treatment of quantum-produced data belong in every access agreement now. Installment Two states the same rule for the island tenancy; this installment extends it to every access contract, hosted or remote, because the asset is the data and not the site.

Takeaway. Labels pay because they have no classical substitute at the frontier, and the party that owns the trained model collects. The tenancy and the procurement decide that party before any result exists.

IV. Sampling and Kernels Survive Only Where Physics Structures the Data

Two channels promise a general improvement to AI and deliver a narrow one. The section states where each survives classical competition and why the general case fails.

Sampling

A quantum sampler draws from distributions that classical samplers provably cannot reproduce at equal cost. The question for AI is whether any such distribution is useful as training data, and the answer so far is one candidate: molecular conformer ensembles, where the physics that makes sampling hard is the physics the downstream model needs to learn. Outside physics, no distribution has been identified that is both hard to sample and informative for a named task.

The classical countermove does not reproduce the distribution. The countermove reproduces the downstream task performance with a task-specific generator, which is cheaper than matching the samples and sufficient for the buyer. The simulation places that countermove within four quarters of any general claim, which is why P-2 states the negative: no quantum-sampled synthetic dataset improves a general AI benchmark at equal compute through 2030.

Kernels and Feature Maps

A quantum circuit can evaluate a similarity measure that is expensive to compute classically, which helps on small datasets with physical structure. The hybrid boson-sampling study is the current state of the evidence: gains against two weak kernels, and no comparison against tuned tree-based or neural classifiers. The 2026 review of quantum kernel methods, Non-Variational Supervised Quantum Kernel Methods: A Review, sets out the conditions a kernel must meet to separate from classical models and finds none of them met on a classical dataset with full data-loading cost included.

Quantum oracle sketching extends the kernel idea in a different direction: it saves memory rather than time. The Google and Caltech result processes data samples one at a time through small quantum operations and matches classical accuracy with far less memory, but data loading still dominates runtime and the result is simulated. The simulation holds it as a contingent extension of the kernel channel rather than a sixth channel, and the activation condition is a result that meets P-1’s qualifying standard on a non-physics dataset running on commercial hardware.

Takeaway. Sampling survives in conformer ensembles and kernels survive on small physics-structured sets. Neither reaches the general AI market within the horizon, and a buyer who hears otherwise should ask for the baseline.

V. Twelve Cognitive Digital Twins Test Which Channel Pays First

The simulation answers one question. Through which channel does quantum first improve an AI system in a result a buyer will pay for, which actor publishes it, and which actor converts it into a market? The horizon runs from October 2026 through 2033, with the United States as forum for buyers and hosts.

The Actor System

Mechanism

Game theory supplies the payoff structure. A hyperscaler earns priority by publishing first and a vendor earns by placing hardware. A classical researcher earns by matching a general claim, and a buyer pays only when the output beats the best classical alternative at equal cost. The equilibrium is a labels result published by a hyperscaler, matched classically at the first system size, and moved to larger systems where the value persists.

Behavioral economics supplies the decision rules. Hosts and lenders respond to the hardware label while the software delivering the gain is invisible to them. Laboratories buy access with no data clause until a replicated result makes the resale value of labels salient, and counties review the island by label regardless of channel. Predictive behavior emerges from the combination.

Scenario Routes

The base route runs labels first: a hyperscaler publishes by 2028 with a stated classical baseline and a laboratory procures labeled data after replication. DARPA names chemistry and clean-hosted islands report labels. Sampling survives in one structured case, and kernels and pipeline optimization stay inside noise.

The alternative route runs classical catch-up: a classical method matches the first labels result at the first system size within two years, the channel’s value shifts two to three years later, and procurement stays at machine access. The tail route runs no channel: nothing produces a buyer-paid result by 2032 and quantum stays a research instrument.

Takeaway. The simulation resolves labels against samples as a question of what already exists, and publisher against host as a question of who earns priority. The Simulation Predictions follow from both.

VI. MindCast Foresight Simulation Predictions

Six Simulation Predictions follow, two primary and four secondary. P marks a Primary Simulation Prediction and S a Secondary Simulation Prediction. Each names an actor and an outcome with a window, a band, a falsifier and a verification source. No Simulation Prediction in this installment has reached its resolution date, so the validated tally stands at zero hits and zero misses.

Simulation Predictions at a Glance

Competing Explanations

Three accounts compete before any result prints.

Labels first. Quantum-computed labels for scientific models produce the first buyer-paid AI result, because the capability exists and the consumer exists. Falsifier: the first qualifying result is a sampling or kernel result.

Classical substitution. Improved classical methods and quantum-inspired software capture every useful gain, and quantum stays an instrument. Falsifier: any qualifying result survives twelve months without a classical match.

Sampling breakthrough. A research group identifies an AI-relevant distribution with provable sampling hardness, and synthetic data becomes the first channel. Falsifier: no such distribution is published with a hardness argument and a downstream task by 2030.

Primary Simulation Predictions

P-1 | Labels Before Samples (62–74%)

Forecast statement: among qualifying results first published on or after October 28, 2026 the first is a materials or chemistry model trained on quantum-computed labels. The first qualifying result is not a model trained on quantum-sampled synthetic data and not a quantum kernel or feature map.

• Qualifying result: a peer-reviewed paper or preprint that meets four conditions. It trains or augments a model with data a quantum processor produced. It reports a gain on a named benchmark or discovery metric against the strongest classical baseline the field uses for that task. It accounts for data-loading, execution and readout cost, and it survives twelve months without a published classical match.

• Band: 62–74%.

• Window: through December 31, 2031.

• Trigger conditions: any result meeting the first three conditions appears; the fourth resolves twelve months later.

• Explicit falsifier: the first result meeting all four conditions is a sampling or kernel result.

• Verification source: arXiv, Nature family journals and hyperscaler and laboratory research publications; the classical-algorithms literature for the survival condition.

P-2 | No General Sampling Advantage for AI (70–82%)

Forecast statement: no peer-reviewed result shows a quantum-sampled synthetic dataset improving a named AI benchmark over the best classical generator at equal compute cost, where the benchmark is not restricted to a physics-structured distribution.

• Band: 70–82%.

• Window: through December 31, 2030.

• Trigger conditions: a vendor or research group claims such a result.

• Explicit falsifier: one peer-reviewed result shows it and survives twelve months without a classical match.

• Verification source: peer-reviewed publications and the classical-algorithms literature.

Secondary Simulation Predictions

S-3 is reported as P(A given B), and the band is void until a laboratory or federal program states authority to procure data deliverables. A replicated labels result is the trigger that moves S-3 inside the window and is not the condition.

Watch Conditions and Break Conditions

Three conditions carry no band until evidence arrives. A published candidate distribution with a hardness argument and a downstream AI task would open the sampling channel to a band. A published model card reporting discovery rate with and without quantum labels would establish the label value ratio Installment Two states as the proposition under test. A quantum oracle sketching result meeting P-1’s qualifying standard on a non-physics dataset would extend the kernel channel.

Three events would break the set. A demonstrated quantum advantage on general machine learning that survives twelve months unmatched would void P-1 and P-2 together. A DARPA finding before 2030 that no team reaches utility scale would push every window past the horizon. A hyperscaler acquiring the clean-hosted anchor tier would remove the island placement case and reduce S-1 to a publication race with no host side.

VII. Strategic Insights by Audience

One question runs under every audience’s decision: which channel will a buyer pay for, and who owns what it produces. Policymakers meet the question in procurement design. Executives meet it in what to sell, counsel in what to allocate and investors in what to credit.

Strategic Posture by Audience

🏛️ Policymakers. Quantum reaches programs and committees as an AI acceleration request, and only the labels channel produces AI output within the horizon. Fund that channel with labeled data as the deliverable, and tie any incentive to a named application class and a stated classical baseline (S-2, S-3).

Procurement is the lever, and a program that writes the first data-deliverable procurement sets the precedent every later contract reads against. No state incentive reaches that decision; the state’s own laboratories and universities hold it. The observable is a procurement notice naming data rather than machine time.

💼 Executives. Hosts should sell the anchor’s quantum role as hub access and research tenancy. Vendors should target chemistry benchmarks with stated classical baselines. Model builders should ingest the first label set above a thousand points and publish the delta (P-1, S-1, S-4).

Installation and placement are separate markets, and executives who conflate them build processors their buyers never visit. The first commercial quantum-AI service delivers labels as data from wherever the machine sits, so a host earns from the tenant’s program and not from the rack. A vendor’s placement offer and a buyer’s remote delivery point in opposite directions, and the buyer’s direction governs.

⚖️ Counsel. The asset is the label set and the model trained on it. Allocate label ownership and derived-model ownership together with publication rights and export-control treatment in every quantum access agreement before the first result exists (S-3).

Counsel are drafting against a precedent that does not yet exist. The first procurement naming data as a deliverable will fix the market’s reading of every earlier access contract, and an agreement silent on output assigns it by whoever holds it when the program ends. A federally funded label set adds a second question, whether a federal data clause overrides the contract, and the agreement should assign who bears that change.

📊 Investors. Credit a channel only after a replicated result survives twelve months without a classical match. Underwrite labels revenue per validated candidate. Carry sampling and kernel revenue at zero until the survival test passes, and pipeline optimization at zero through the horizon (P-1, P-2).

Classical catch-up is the risk that distinguishes the channels. A sampling or kernel claim that gets matched loses its revenue line, while a labels result that gets matched moves to larger systems and keeps it. The survival test separates the two, and the three-site hub contract from Installment Two is the first observable that converts labels revenue from option value into contracted value.

Where Audience Interests Collide

Hyperscalers and laboratories want different firsts. A hyperscaler wants the first publication and will withhold the label set to protect its model lead, and a laboratory wants the label set for its own programs. The first procurement naming data as a deliverable resolves the collision in the laboratory’s favor, and S-3 reads that notice as the evidence.

Vendors and buyers want different deliveries. A vendor earns on each placed processor and a buyer takes labels remotely, so the vendor’s placement offer competes with the buyer’s procurement for the host’s decision. S-4 resolves which wins, and a host that waits for the buyer keeps the option the vendor’s offer would close.

Classical researchers and quantum vendors want different baselines. A vendor publishes against the baseline that makes the gain visible, and a classical researcher publishes the baseline that removes it. P-1’s qualifying standard adopts the researcher’s baseline, which is why vendor results against linear kernels do not enter the population.

Takeaway. Each audience’s best move follows from treating the label set as the asset, and the collisions resolve on dates the Simulation Predictions already carry.

VIII. Risk Mitigation by Simulation Prediction

Every entry below pairs a Simulation Prediction with the exposure it creates for one audience, the actions that audience can take alone and the exposure that survives. Severity and probability run on separate axes throughout.

Mitigating Actions by Time Gate

• Now to 30 days: strip every quantum-AI claim from pending filings and investor documents that lacks a stated classical baseline.

• 30 to 90 days: add label, model and publication ownership clauses to every quantum access agreement in negotiation.

• 3 to 12 months: write the first quantum procurement or tenancy for a chemistry or materials program with a named model and a data deliverable.

• Beyond 12 months: keep the anchor’s reserved circuit sized for an island and leave the processor unordered until a named tenant workload requires it on site.

Exposure, Actions and Residual for Each Simulation Prediction

The actions are analytic options for each audience and not legal, investment or fiduciary advice to any party.

Entry 1 · 💼 Clean-hosted anchor executives · P-1 (62–74%).

• Exposure: a host that specifies its island or its tenancy for a sampling campaign or a general AI buyer loses the one tenant class the island can serve, and carries the island’s fixed cost for the quarters until a chemistry tenant replaces it. Severity: high. Probability: moderate.

• Mitigating actions: the commercial lead writes the first tenancy for a chemistry or materials program with a named model, before the island is specified. The engineering lead sizes the island’s coupling for active-learning cycles in minutes rather than sampling throughput. The finance lead models revenue per validated candidate only.

• Residual exposure: the first tenant’s model may underperform its classical baseline on the chosen target.

Entry 2 · 📊 Investors in quantum vendors and quantum-equipped sites · P-2 (70–82%).

• Exposure: a valuation that credits a vendor’s sampling-based AI claim carries the claim’s revenue until a classical match removes it, which the simulation places within four quarters of publication. Severity: moderate. Probability: high.

• Mitigating actions: the diligence lead requires a stated classical baseline and a twelve-month survival test before crediting any sampling or kernel revenue. The credit committee credits labels revenue only after a second group replicates.

• Residual exposure: one physics-structured sampling case may survive and carry value the test excludes.

Entry 3 · ⚖️ Counsel to laboratories, hosts and model builders · S-3 (40–52% given authority).

• Exposure: an access contract silent on label and model ownership assigns the channel’s value by default to whichever party holds the output when the first procurement names data as a deliverable, and the other party renegotiates under time pressure. Severity: high. Probability: moderate.

• Mitigating actions: transaction counsel allocates label ownership, derived-model ownership and publication rights in every access contract now. Regulatory counsel documents the client’s assessment of export-control treatment for quantum-produced data before the first cross-border transfer.

• Residual exposure: a federal procurement clause can override contractual allocation for federally funded labels.

Entry 4 · 🏛️ Policymakers and research-program officers · S-2 (64–76%).

• Exposure: a program that funds quantum access for optimization or general AI ahead of chemistry spends a budget cycle on the class DARPA’s verification is least likely to name, and loses the procurement precedent to a program that funded chemistry. Severity: moderate. Probability: moderate.

• Mitigating actions: program staff write the first quantum access procurement for a chemistry or materials program with a data deliverable, before the first final-stage report. Legislative staff condition any state quantum incentive on the application the program names and the classical comparison it commits to.

• Residual exposure: DARPA may name a class the program did not fund, or name none.

Entry 5 · 💼 Model builders and enterprise chemistry buyers · S-1 (55–67%).

• Exposure: a buyer who waits for a clean-hosted island to produce the first labels model licenses two to three years later than one who licenses from the first hyperscaler publication. Severity: moderate. Probability: moderate.

• Mitigating actions: the research lead licenses or replicates the first published labels model regardless of its host, by the quarter after publication. The procurement lead writes host-neutral data clauses so a later island tenancy can substitute.

• Residual exposure: a hyperscaler can withhold the label set and license only the model.

Entry 6 · 💼 Clean-hosted anchor executives · S-4 (64–76%).

• Exposure: a host that justifies an island on the expectation that a quantum-AI product creates local processor demand carries the island’s fixed cost while the product’s buyers take delivery remotely from a shared hub. Severity: high. Probability: moderate.

• Mitigating actions: the commercial lead sells the anchor’s quantum role as hub access and tenancy for the research program before any island quote. The engineering lead keeps the reserved circuit and defers the processor until a tenant names a workload that needs local coupling.

• Residual exposure: a warm-modality processor can make local ownership ordinary and move demand back to sites.

Takeaway. Treating the label set as the asset protects every audience under both outcomes of P-1, because a buyer still pays for labels if sampling never arrives and still takes delivery remotely if it does.

IX. Implications Run From the First Publication to the First Procurement

Hyperscalers capture publication priority on the labels channel, because they hold the hardware queue and the models that consume labels (S-1). Laboratories capture the channel’s market by procuring data as a deliverable after replication, and the procurement activates the multi-site buyer that forms Installment Two’s hub (S-3). Clean-hosted anchors capture tenancy and hub access, and nothing more, because the product delivers remotely (S-4).

Classical catch-up sorts the channels. Sampling and kernel claims lose their revenue line when matched, and labels results move to larger systems and keep theirs (P-1, P-2). Vendors that target chemistry benchmarks with stated baselines keep access revenue; vendors that publish against weak baselines lose lender credit for every later claim.

If P-1 misses on a sampling result, an AI-relevant distribution existed that the simulation did not find, and the sampling channel moves to first. If P-1 misses on a kernel result, the classical match lagged, and the oracle-sketching extension gains weight. If S-4 misses, a warm-modality processor made local ownership ordinary, and the island route in Installment Two gains weight against the hub.

Takeaway. Publication, procurement and placement are three decisions by three actors. A miss at any one moves the date without changing the channel that pays.

What to Watch: The First Label Set, the First Procurement and the First Service

The dominant fork is whether the first qualifying result is a labels result or a sampling result (P-1).

• 2027 and 2028. A label set above a thousand points on a correlated system released to benchmark groups; the first qualifying paper with a stated classical baseline.

• Each DARPA final-stage report. The named application class moves S-2 and the vendors’ benchmark targets.

• December 31, 2030. P-2 resolves against peer-reviewed publications; S-3 resolves against procurement notices, given authority.

• December 31, 2031. P-1 and S-1 resolve against research publications and the survival condition.

• December 31, 2033. S-2 resolves against DARPA’s report; S-4 resolves against the first service announcement naming its quantum source.

Verification sources by Simulation Prediction.

• Resolves P-1 and S-1: arXiv, Nature family journals and hyperscaler research publications reviewed quarterly, with the classical-algorithms literature for the survival condition.

• Resolves P-2: peer-reviewed publications claiming sampling-based AI gains, reviewed at each claim.

• Resolves S-2: DARPA Quantum Benchmarking Initiative reports, reviewed as they appear.

• Resolves S-3: federal and laboratory procurement notices, reviewed quarterly.

• Resolves S-4: service announcements naming the quantum source, reviewed quarterly.

• Moves watch conditions: candidate hard distributions with downstream tasks, model cards with and without quantum labels and oracle-sketching results on commercial hardware, reviewed as they appear.

Conclusion: Fund the Labels, Own the Model and Take Delivery Remotely

Quantum improves AI first through labels for scientific models on correlated matter. On the base route the first qualifying publication arrives by 2028 with licensing by 2029, and the alternative route delays both by two to three years when a classical method matches the first result. Sampling survives in one physics-structured case, kernels and pipeline optimization stay inside noise, and quantum-inspired software delivers today on hardware the anchor already owns. The party that owns the model trained on the labels collects, and the buyer takes delivery from wherever the machine sits.

Stakeholders should act now on what they control: the baseline standard in every claim they credit, the ownership clauses in every access agreement they sign, and the application class in every procurement they write. Stakeholders can wait to learn whether a classical method matches the first labels result and whether DARPA names chemistry first.

The decisive signal is the first qualifying result published after October 28, 2026. If it is a labels result for a materials or chemistry model, P-1 holds and the labels route governs the bridge between the anchor and its quantum tenant. If it is a sampling or kernel result, P-1 misses and the model expects the channel order to invert while the anchor’s classical case stands.

Working With MindCast

Public analysis cannot tell a specific laboratory, host or model builder which system size its first label set should target or what its access agreement should assign. Nor can it say whether the buyer will take delivery remotely. The answers depend on the program’s mission systems, the counterparty’s contracting position and the buyer’s integration cost. Program inputs stay confidential.

MindCast builds Cognitive Digital Twins of the program, its vendor, its model-building partner and its buyer. The engagement ingests the organization’s own decision history and reruns the channel contest and the ownership terms against its actual options.

Commissioned scope adds complete probability distributions, P10/P50/P90 ranges, organization-specific calibration and complete scenario trees. The scope also adds the net quantum-AI value test run on the organization’s own classical baseline and delivery costs, and red-team tests of vendor baseline claims.

Three engagements map to this installment’s readers. A Channel Diagnostic for a laboratory or model builder: one program, one target system and one channel decision (P-1, S-3). An Access Terms Review for counsel tests label, model and publication ownership together with export-control treatment against vendor and buyer responses (S-3). A Procurement Brief for a program office: the data-deliverable procurement run against vendor and laboratory Cognitive Digital Twins (S-2, S-3).

Contact mcai@mindcast-ai.com or visit www.mindcast-ai-simulation.com to arrange one.

Appendix A. MindCast Works

Companions

Regional Clean Compute Anchors, How Much Compute Each Clean Energy Source Yields and Where Washington Builds the First One (Installment One). Sets the anchor configuration and its reserved island circuit. Supplies the host whose quantum role this installment limits to tenancy and hub access.

What Quantum Computing Contributes to Clean Compute Clusters, Tenant First, Training Data Second and Not the Processor Through 2032 (Installment Two). States the inference wall, the training-data loop and the hub-versus-island contest. Supplies the boundary this installment takes as given and the tenancy rule it extends to every access agreement.

How Quantum Technology Changes the AI Data Center Energy, Compute and Environmental Calculus (October 2026). The quantum series overview. Supplies the finding that quantum shifts constraints rather than relieving them, which the five-channel ranking applies to AI output.

Funded Solutions That Can Help Win Quantum-Coupled AI Data Centers Approvable (October 2026). Tests 31 measures across three project types. Supplies the separation of quantum-inspired software from quantum hardware that places the fifth channel.

The series umbrella and the market

AI Data Center Authorization + Clean Energy, the Mid-Scale Squeeze and the Specialized Compute Landscape(September 2026). The series umbrella. Supplies the Mid-Scale Squeeze that defines the enterprise chemistry buyer’s tier.

Where the Clean Energy Host Advantage Holds and Where It Breaks in AI Data Centers (October 2026). Cover letter on host sites. Supplies the hyperscaler hosting forecast behind S-1’s publication race.

AI Datacenter Edge Computing: Ship the Workload Not the Power (September 2025). The founding principle of moving compute toward energy. Supplies the siting logic that S-4 inverts for labels, which move toward the buyer instead.

Appendix B. External Sources

• DARPA. “Four More Teams Enter Quantum Benchmarking Initiative Final Stage.” October 7, 2026. The six final-stage teams and the 2033 utility-scale question.

• IBM. “IBM Sets the Course to Build World’s First Large-Scale, Fault-Tolerant Quantum Computer.” June 2025. The qLDPC architecture and about 200 logical qubits by 2029.

• Google Quantum AI. “Quantum Error Correction Below the Surface Code Threshold.” Nature, December 2024. Logical error falling with code size.

• Tang, E. “A Quantum-Inspired Classical Algorithm for Recommendation Systems.” 2018. The dequantization line of work that bounds the sampling and kernel channels.

• Sharifian and Bayat. Hybrid Boson Sampling-Neural Network Architecture for Enhanced Classification arXiv preprint of October 2025 with a 2026 journal version. Gains against linear and sigmoid kernels on four datasets; no tuned classical baseline.

• Non-Variational Supervised Quantum Kernel Methods: A Review arXiv, April 2026. The conditions a quantum kernel must meet to separate from classical models.

• Zhao and six co-authors at Google Quantum AI and Caltech. Exponential Quantum Advantage in Processing Massive Classical Data arXiv 2604.07639 of April 2026 (revised October 1). Quantum oracle sketching; memory advantage in simulation on sentiment and single-cell data under 60 logical qubits.

• Equinix and Diraq. “Diraq and Equinix to Deploy Australia’s First Quantum Computer Inside a Data Centre.”September 1, 2026. Remote-access precedent inside a production data center.

• IonQ. “IonQ to Advance Quantum Supercomputing by Bringing First QPU to NVIDIA Accelerated Quantum Research Center.” September 23, 2026. Classical-quantum coupling through NVQLink.

Companion works: Regional Clean Compute Anchors · How Quantum Improves AI · How Quantum Technology Changes the AI Data Center Energy, Compute and Environmental Calculus · Funded Solutions That Can Help Win Quantum-Coupled AI Data Centers Approvable

Why This Installment Matters to Each Audience

Each audience will meet quantum equipment in a different form, and the installment states what each should conclude before the first request arrives.

🏛️ Policymakers. Quantum equipment reaches committees and commissions first as a continuity request and only later as a research asset. The installment supplies the functional test that separates the two, and it states why a label-based protected-load rule invites every later technology to claim it. Legislators who want the hub in their state learn which lever forms one: a multi-site research procurement rather than a tax incentive.

💼 Executives. Hosts and vendors face a hub-or-island decision in the 2027 hosting cycles, before any utility-scale evidence exists. The installment states the break-even in sites served at each utilization level and the installation-speed asymmetry that lets islands win a race a hub would win on economics. Executives who install an island first carry its fixed cost for the quarters until a tenant arrives.

⚖️ Counsel. The island’s highest-value output is data no existing tenancy template allocates: quantum-produced labels and the models trained on them. The installment identifies the ownership, publication and export-control terms that must exist before the first label does, and the continuity duties the hosting agreement must carry. Counsel who treat the island as ordinary colocation leave the value of every later discovery unassigned.

📊 Investors. Quantum readiness carries option value and zero near-term capacity value, and underwriting cases routinely conflate the two. The installment supplies the utilization assumption, the per-candidate revenue unit and the three-site hub test that separate them. Investors who credit an island with hub-level utilization overstate annual margin by the island’s fixed cost.

Takeaway. The installment gives each audience the unit its quantum decision should run in, and Sections IX and X state what to conclude and what to do with it.

I. Quantum Joins the Cluster as a Tenant First and a Data Source Second

Quantum computing does not make artificial intelligence compute more efficient within this paper’s horizon. No roadmap from IBM, Google, NVIDIA or DARPA claims it will by 2032. Quantum raises research productivity in chemistry and materials, measured in time to a validated candidate. It joins a regional clean compute anchor as a research tenant first and as a source of training data second.

The training data lets scientific models reason about matter whose ground truth classical simulation cannot supply, and that is the whole of quantum’s contribution within this paper’s horizon.

Thesis. A quantum island under one megawatt sits on a firm-clean anchor of 8 to 20 megawatts and serves one workload class: synthesis and simulation for materials, chemistry and optimization. Under the stated assumptions the island covers its fixed cost only as a shared hub serving six or more sites or under a public research tenancy, and never as a local installation at ten to twenty percent utilization. Game Theory favors the shared hub on utilization, and Behavioral Economics predicts that hosts will overbuild local islands anyway.

A research buyer whose programs span several sites is the actor most likely to force the hub, and the paper states that mechanism as conditional because no laboratory has yet stated a hosting position.

Why now. On October 7, 2026 the Defense Advanced Research Projects Agency (DARPA) moved four more companies into the final stage of its Quantum Benchmarking Initiative. Atom Computing, Diraq, IBM and IonQ join Microsoft and PsiQuantum there. The initiative asks whether any approach reaches utility scale by 2033.

Equinix and Diraq announced on September 1, 2026 a planned deployment of a quantum computer drawing under 20 kilowatts inside a production data center. IonQ announced on September 23, 2026 that its Superion 256 will couple to an NVIDIA GB200 NVL72 system through NVQLink, with installation in 2027. Energy Northwest’s reactor project near Richland plans construction by 2030, and the paper models operation in the mid-2030s, after DARPA’s 2033 answer. Seattle holds a quantum cohort of its own: Microsoft, Pacific Northwest National Laboratory and the University of Washington run quantum programs within the latency radius, and IonQ manufactures in Bothell.

What distinguishes the MindCast approach. Vendors publish roadmaps and physicists publish benchmarks. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through MindCast AI Proprietary Cognitive Digital Twin Foresight Simulations (MP CDT FS). Each Cognitive Digital Twin (CDT) models how a named institution decides as evidence and rules change, and Section VII names the eleven this installment runs.

Behavioral Economics supplies the decision rules: “quantum-ready” is salient at the county and the lender while utilization is not. Game Theory supplies the payoff structure: the host and the buyers gain from a shared hub while the vendor’s placement revenue and faster installation pull toward islands. Predictive simulations emerge from the combination.

What the paper does. Section II takes the anchor as given and corrects the assumption that quantum speeds compute. Section III states what quantum contributes by tier and date. Section IV names the training-data loop and holds planning-loop optimization as a watch condition.

Section V ranks the innovations that would enlarge quantum’s role. Section VI states the economics once quantum joins and phases the island across years three and five. Section VII states the simulation architecture, Section VIII carries the MindCast Foresight Simulation Predictions. Section IX states the strategic insights for each audience, and Section X carries the Risk Mitigation entries.

Simulation Prediction highlights. Five Simulation Predictions in Section VIII.

• P-1. Given a research buyer whose programs span three or more sites: the first commercial quantum hub on a clean energy host serves at least three remote sites at job level before any second clean host installs a local island. Window through December 31, 2031. 48–60% given the buyer.

• P-2. The first clean-hosted quantum island reports a chemistry or materials model trained on its own labels before it reports any energy relief or capacity metric, through December 31, 2032. 44–56%.

• S-1. No published benchmark shows a quantum or hybrid solver beating the incumbent classical solver at equal solve time on a data center dispatch problem, through December 31, 2030. 78–88%.

Stakeholder readout.

• 🏛️ Washington legislators. Quantum equipment adds one continuity objection and no energy relief. Tie any protected-load or critical-load treatment to function and recovery time rather than to the quantum label, and align the state’s laboratory and university hosting authority with the multi-site procurement that forms a hub (P-1, S-2).

• 💼 Seattle quantum cohort and clean energy hosts. IonQ’s Bothell manufacturing site and Microsoft’s Redmond quantum program sit within the latency radius of the first Washington anchors. Join or form a shared hub before installing an island, and sell the island on per-candidate research output with a named laboratory partner (P-1, P-2).

• ⚖️ Counsel to the quantum cohort. Training labels and the models trained on them are the island’s highest-value output, and research access is the other revenue channel. Allocate ownership and publication rights and export-control treatment in the tenancy before the first label exists. Write the ride-through duty and protected reserve into the hosting agreement (P-2, S-3).

• 📊 Investors. Give quantum zero capacity credit and underwrite it at ten to twenty percent utilization until a hub serves three sites. Value output per validated candidate once a trained model exists (P-1, P-2).

II. The Anchor Taken as Given, and the Assumption to Remove First

Installment One is Regional Clean Compute Anchors, How Much Compute Each Clean Energy Source Yields and Where Washington Builds the First One. It sets the anchor: a firm-clean site of 8 to 20 megawatts hosting regional enterprise inference, paired with solar and wind modules that carry movable work, approved as an accessory use at an energy site. The anchor reserves a protected circuit for a quantum island under one megawatt. The question for this installment is what the island contributes, and the answer begins by removing the assumption most readers bring.

Why Quantum Does Not Accelerate Inference or Training

An inference pass is dense matrix multiplication over billions of classical parameters and classical input tokens. Three walls block a quantum machine from that job. Loading classical data into a quantum state costs about as much as the computation it would accelerate, so the input step consumes the speedup. Reading the answer out collapses the state, so a single pass yields a few bits rather than a vector of activations.

Gate speeds run in microseconds against nanoseconds for silicon, and error correction multiplies every logical operation by hundreds of physical ones. The early quantum machine learning results claimed exponential speedups for recommendation and clustering. Classical researchers matched them with new classical algorithms, a line of work known as dequantization, and the claimed advantage disappeared. Grover’s quadratic speedup for search is real on paper and vanishes under error-correction overhead on actual hardware.

IBM, Google and NVIDIA position quantum beside GPUs rather than in place of them for exactly this reason. A one-megawatt quantum island beside a 20-megawatt GPU hub adds zero inference capacity, and a reader who expects otherwise is importing the Moore’s Law frame into a machine that does not run on it.

Takeaway. Quantum raises research productivity measured in time to candidate. It does not raise compute productivity measured in tokens or floating-point operations.

III. What Quantum Contributes, by Tier and Date

Quantum’s contribution arrives in three tiers, and only the first two land within the cluster’s horizon. Each tier names what the machine produces and the unit in which productivity rises.

Tier One, Now Through 2029: A Scientific Instrument and Quantum-Inspired Software

Tier one separates into what is demonstrated and what is projected. Demonstrated: chemistry beyond exact classical reach already runs at 50 to 100 qubits. IBM’s sample-based quantum diagonalization pairs Heron processors with classical supercomputers on molecular ground-state problems at sizes exact classical methods cannot store.

IBM’s demonstrated output is energies and reaction pathways. The productivity gain the paper tests is a compression of laboratory screening time in catalysis and battery chemistry, and no published result has yet measured it on an industrial molecule.

Error correction has crossed its threshold. Google’s Willow result in December 2024 showed logical error falling as code size grew, the precondition for every later tier. Quantinuum’s Helios runs near 100 trapped-ion qubits, and IBM’s roadmapreaches about 200 logical qubits by 2029.

Quantum-inspired classical algorithms deliver gains today. Tensor networks and annealing-derived heuristics improve scheduling, logistics and portfolio optimization, and they run on ordinary GPUs. Most quantum productivity cited in corporate materials is this software, which needs no quantum hardware. MindCast’s companion Funded Solutions That Can Help Win Quantum-Coupled AI Data Centers Approvable already separates its value from quantum hardware, and the separation keeps every software gain out of the island’s case.

Projected: useful advantage on a named industrial molecule has not appeared, and compressed screening time is the expected gain once it does.

Takeaway. Tier one quantum is a demonstrated laboratory instrument and a demonstrated software method, and the productivity gain is projected.

Tier Two, 2029 to 2033: Error-Corrected Machines at 100 to 1,000 Logical Qubits

Materials and drug discovery reach useful fidelity at this tier: electrolyte design, nitrogen-fixation catalysts and protein-ligand binding energies. The classical cost of these simulations grows exponentially with electron count, and the quantum cost grows polynomially. The productivity unit is time to a validated candidate and the number of candidates a research budget can screen. DARPA’s benchmarking initiative asks whether any of its six final-stage teams reaches utility scale by 2033.

Narrow optimization advantage appears in problems with the right structure. The advantage is modest and smaller than annealing vendors claim. Quantum-generated training data becomes the durable product, and Section IV states the loop.

Takeaway. Tier two turns the instrument into a data source.

Tier Three, After 2033: Cryptographically Relevant Machines

Factoring RSA-2048 needs under a million noisy qubits running about a week on 2025 estimates, down from twenty million in 2019. The implication for every data center is defensive: post-quantum cryptography migration, not compute. The cluster inherits that obligation on the same schedule as every other facility.

Takeaway. Tier three changes security budgets rather than compute budgets.

IV. Quantum-Labeled Training Data Is the Real Role

The island’s durable contribution is data that only a quantum machine can produce, consumed by models that run classically at scale. The section states the loop, its volume constraint and the one optimization use that stays a watch condition.

The Training-Data Loop

The loop below is a projected workload. No published materials model yet carries quantum-computed labels.

Train a model on data that only a quantum machine can produce, then run that model on GPUs at scale. Classical simulation of electronic structure is approximate once electrons correlate strongly. Catalysts, battery electrolytes, superconductors and reaction dynamics all live in that regime. A fault-tolerant quantum machine computes those energies and dynamics to chemical accuracy at sizes classical methods cannot reach.

The outputs become labeled training data for neural network potentials and materials foundation models of the kind Google, Microsoft and Meta already train on classical density-functional data. Discovery runs as an active-learning loop. The model proposes candidates and the quantum machine validates the handful worth the expense. The validated labels retrain the model and the next proposals improve, with the quantum island as the oracle and the GPU host as the proposer and learner.

Three quantities decide the loop’s value: whether quantum-generated labels are available at the horizon, their accuracy against the best classical method and their economic value per validated candidate. The comparison that decides them is equal useful output against the best classical alternative. Preparation and error correction sit inside the quantum cost, and so do measurement, verification and queue time.

Tight coupling matters here for the first time in the cluster model, because each cycle moves proposals and labels between the island and the GPU host in minutes, and a remote hub adds queue time to every iteration. Synthesis in Installment One’s workload envelope is this loop.

Takeaway. Quantum supplies the labels that let scientific models reason about correlated matter, and the island becomes an input to a product the anchor already sells.

What “Reasoning” Means Here and the Volume Constraint

The model gains the ability to reason about physical systems whose ground truth was previously unavailable. A language model gains nothing, because quantum data says nothing about text, code or planning. The claim is a new tier of scientific artificial intelligence rather than of machine reasoning in general.

Quantum machines produce few samples slowly, so the value sits in high-cost labels, not bulk data. The economic proposition the paper tests is that a thousand chemically accurate energies on the right molecules are worth more to a scientific model than a million approximate ones, and that ratio would justify a one-megawatt island beside a 20-megawatt GPU hub. The observable is a published materials model whose training set carries quantum-computed labels and whose validated discovery rate beats the classical-only baseline.

Takeaway. A reader who hears “quantum makes models reason better” will apply it to chatbots and be wrong; the exact claim is scientific models and correlated matter.

Planning-Loop Optimization Is a Watch Condition, Not a Benefit

Inference has two loops around it. The control loop routes each request and places prefill and decode in microseconds to milliseconds. The planning loop decides which models sit at which site, how much power goes to inference each hour and when storage discharges, in minutes to a day. Quantum cannot enter the control loop, because a quantum job carries queue, compile and readout time measured in seconds to minutes.

The planning loop is a combinatorial problem that annealers and hybrid solvers target. Classical solvers already produce near-optimal plans for problems this size in milliseconds. The gap between near-optimal and optimal on a daily dispatch is a modeled range of one to three percent of energy cost or utilization, which the simulation tests instead of assuming, and every published quantum advantage on scheduling has been matched by classical heuristics on review. S-1 states the watch condition.

Takeaway. Quantum can improve the plan that shapes inference by low single digits, and only after it beats a classical solver.

V. Innovations That Would Enlarge Quantum’s Role

Six innovations would change how much the island contributes, ranked by their effect on the cluster, each with the observable that marks its arrival.

1. Low-overhead error correction. IBM’s qLDPC codes cut physical qubits per logical qubit by up to 90 percent. Every reduction shrinks the island’s refrigeration, power and cryogen burden, which is the entire approval cost of local quantum. Observable: a vendor publishes a logical-qubit count with a physical-to-logical ratio under 100.

2. Room-temperature and warm modalities. Neutral-atom, photonic and trapped-ion systems run without dilution refrigeration or helium-3. A room-temperature processor is an ordinary rack with a laser table, and the continuity, cryogen and hub land-use objections disappear. Observable: a colocation host installs a quantum processor with no cryogenic plant and no ride-through beyond standard uninterruptible power.

3. Classical-quantum interconnect. NVQLink-class links and cryogenic control electronics push real-time decoding onto GPUs and extend the coupling radius from meters toward a campus. Observable: a published decode loop under a millisecond across a building.

4. A demonstrated advantage on an industrial chemistry problem. The single event that converts the synthesis tenant from research budget to commercial contract. Observable: a reproducible result on a named catalyst or electrolyte that a classical supercomputer cannot match at equal fidelity.

5. Quantum networking between islands. Entanglement distribution lets several small islands act as one machine. Observable: two-site logical operations over metro fiber.

6. Efficient quantum memory and data loading. The only innovation that would touch inference, because it would let a quantum machine ingest classical vectors without the loading penalty. No credible roadmap exists, and the paper names it as the reason quantum stays off the GPU side.

The first three innovations shrink the island until it fits a module. The fourth creates the customer and the fifth expands the hub. The sixth is not coming.

Takeaway. The cluster model is robust to every innovation on the list except the sixth, so the paper states quantum’s role once without waiting on the roadmap.

VI. The Economics Once Quantum Joins, and the Phases in Which It Arrives

The island lands a fixed charge on a small margin and earns through three channels, none of them a GPU-hour product. The section states the cost, the revenue, the equilibrium and the two phases in which quantum arrives.

Cost Lands on a Small Margin

MindCast’s funded-solutions companion puts the quantum delta at about $2.5 million a year at a hub, before specialized staffing and integration software that add another $1 million to $2 million. The island’s fixed cost therefore runs $3.5 million to $4.5 million a year. A 20-megawatt anchor earns a classical operating margin that the paper sets at $5 million to $10 million as a stated assumption, so the island consumes between 35 and 90 percent of it across the endpoints before producing anything. Protected energy stays trivial: under one megawatt of load and about 8 megawatt-hours of load energy for eight hours at the one-megawatt ceiling, before conversion losses and reserve margin set the installed battery above it.

The island also converts part of the anchor from curtailable load into firm demand. The converted megawatt leaves the flexibility programs that earned the site its approval route, and the companion finds approvability falling from 90 to 87 percent at hubs. Quantum spends some of the advantage that justified the anchor.

Takeaway. The island costs the anchor between a third and nearly all of its margin, and three points of approvability, before it produces anything.

Revenue Arrives Through Three Channels

Per-result contracts for synthesis in materials, chemistry and logistics charge for the delivered answer, not the hour. Hub access sold to many remote sites at job level is the lever that matters. Anchor tenancy from a Department of Energy program, a sovereign buyer or a national laboratory underwrites the hub and gives the firm clean host a tenant that cannot leave.

The break-even arithmetic rests on three stated inputs. With fixed cost F of $3.5 million to $4.5 million a year, access revenue R per remote site at full utilization and utilization u, the hub needs F divided by R times u sites. Utilization here means billable quantum processing time, not reserved access or system availability.

At R of $1 million and u of 0.6 the hub needs six to eight sites. At u of 0.3 it needs twelve to fifteen. At u of 0.15, the range a local island reaches, it needs 24 to 30 and no region supplies them. The six-to-eight figure is a sensitivity case and not a deployment threshold.

The simulation carries enterprise synthesis contracts at $0.5 million to $2 million a year as a conditional scenario with no benchmarked contract behind it, so the per-result channel alone needs two to nine contracts or one public anchor to clear the fixed cost. Once the training-data loop runs, the island stops being a tenant and becomes an input to a product the anchor already sells. The synthesis contracts then charge per validated candidate, and the research tenant becomes the first customer rather than the subsidy.

Takeaway. Under the stated assumptions the island pays only as a shared hub or under a public tenancy, and the training-data loop is what converts the tenancy into a product.

The Equilibrium and the Error

Game theory favors the shared hub over distributed islands, because the host and the buyers gain from utilization and lose from duplication. A vendor’s local-placement incentives pull the other way and can reverse the outcome, which is why P-1 carries a band below two in three even with a buyer in place. Behavioral economics predicts the host’s error: islands get overbuilt because “quantum-ready” is salient at the county and the lender while utilization is not. Predictive behavior emerges from the combination, and P-1 states the result.

The actor that converts the equilibrium into an outcome is a research buyer whose programs span sites. Multi-site procurement is the one demand a single-host island cannot satisfy, so a laboratory or federal program that conditions its contract on serving three or more sites forces the vendor to a hub offer. No laboratory has stated a hosting position, so the mechanism is conditional and P-1 carries the condition. A shared hub is an organizational innovation before it is a computational one, because it pools equipment and improves utilization without assuming quantum replaces anything classical.

Takeaway. The equilibrium is a few hubs, the multi-site research buyer is the actor that reaches it, and the observed behavior will include islands that never do.

Two Phases

Year three, 2029. On the modeled base route, one shared quantum hub attaches to a geothermal or hydro anchor and serves the region at job level. The route holds only where a multi-site research buyer contracts for access, and P-1 in Section VIII carries that condition. DARPA’s benchmarking stage and the first coupled installations decide whether the hub is research or product.

The first approval question for the hub is continuity for the protected island, which MindCast’s quantum overview How Quantum Technology Changes the AI Data Center Energy, Compute and Environmental Calculus expects to require backup rather than a curtailment exemption.

Year five, 2031 to 2032. A local protected island sits at the anchor, conditional on utility-scale evidence that the overview places after 2030. The island arrives on hydro or geothermal firm power, because the Richland reactor plans construction by 2030 and operation in the mid-2030s. Synthesis becomes a commercial workload through the training-data loop in this phase.

Year eight and after, mid-2030s. The first small modular reactor module firms an anchor at three times the hydro scale. The Richland corridor then holds a reactor and hydro beside a national laboratory and a quantum tenant under four owners.

Takeaway. On the base route quantum joins at year three as a shared hub and at year five as a local island, and the reactor follows in the mid-2030s rather than on the same calendar.

VII. Eleven Cognitive Digital Twins Test Whether a Hub or an Island Forms First

The simulation answers one question. Does the first clean-hosted quantum installation form as a shared hub serving many sites or as a local island, and which workload does it report first? The horizon runs from October 2026 through 2032, with the United States as the forum and non-U.S. hosting as a comparison case.

The Actor System

Mechanism and Scenario Routes

Game theory supplies the payoff structure. A shared hub pays every party through utilization while a local island pays the vendor through placement and costs the host through continuity.

Behavioral economics supplies the decision rules. Hosts anchor on the salient label and counties discount claims they cannot measure. Authorities decide under docket pressure with decision noise in place of perfect optimization, and predictive behavior emerges from the combination.

Among configuration outcomes, the simulation places the shared-hub route as the base: a few quantum hubs on firm clean anchors serve many sites. Distributed local islands sit second, driven by vendor placement incentives, label salience and one timing fact: an island installs in two quarters while a hub needs about six to serve three sites. Campus-only quantum sits as the tail, where hyperscalers absorb every island and no anchor forms.

Among workload outcomes, the training-data loop reported first is the base route. An energy or efficiency metric reported first sits second, which signals a host that sold the island on claims the quantum overview expects to fail. No workload disclosure at all sits as the tail.

Takeaway. The simulation resolves hub against island as a timing race the buyer narrows but does not win, and labels against metrics as a sale-framing contest, and the Simulation Predictions follow from both.

VIII. MindCast Foresight Simulation Predictions

Five Simulation Predictions follow, two primary and three secondary. P marks a Primary Simulation Prediction and S a Secondary Simulation Prediction. Each names an actor and an outcome with a window, a band, a falsifier and a verification source. No Simulation Prediction in this installment has reached its resolution date, so the validated tally stands at zero hits and zero misses.

Simulation Predictions at a Glance

Competing Explanations

Three accounts compete before any result prints.

Shared hub. A multi-site research buyer forces vendors to hub terms and one hub serves many sites. Falsifier: a second clean-hosted island installs before any hub serves three sites, with a qualifying buyer in place.

Distributed islands. Vendor placement revenue and two-quarter installation lead times put islands at several hosts before any hub ramps. Falsifier: the first hub reports three sites before any second island installs.

Campus only. Hyperscalers absorb every island and no clean-hosted installation forms. Falsifier: any clean-hosted quantum installation by 2031.

The simulation favors the first account at 50 to 60 percent of the weight, the second at 25 to 35 and the third at 10 to 15.

Primary Simulation Predictions

P-1 | Shared Hub Before Distributed Islands, Given a Multi-Site Research Buyer (48–60%)

Forecast statement: P(the first commercial quantum hub on a clean energy host serves at least three remote sites at job level before any second clean energy host installs a local quantum island, given that a research buyer whose programs span three or more sites contracts for clean-hosted quantum access).

• Band: 48–60%, conditional; void if no qualifying buyer contracts by the window. The buyer’s contract pulls the vendor toward a hub, and a second island at another host can still install before the hub serves three sites.

• Window: through December 31, 2031.

• Trigger conditions: a quantum vendor announces a hosted installation on a geothermal, hydro or nuclear site.

• Explicit falsifier: a second clean-hosted local island installs before the first hub reports three remote sites.

• Verification source: vendor and host releases and hosting specifications.

P-2 | Training Labels Before Energy Metrics (44–56%)

Forecast statement: the first clean-hosted quantum island reports a chemistry or materials model trained on labels the island produced before it reports any energy relief, capacity or efficiency metric.

• Band: 44–56%.

• Window: through December 31, 2032.

• Trigger conditions: a clean host or its tenant publishes a workload disclosure.

• Explicit falsifier: the first disclosure reports an energy or capacity metric with no training use.

• Verification source: host, vendor and laboratory publications.

• Scope: P-2 tests commercial positioning, meaning what a host chooses to report first. It does not test whether quantum labels improve scientific productivity, which the watch condition on a quantum-labeled model beating its classical baseline carries separately.

Secondary Simulation Predictions

Watch Conditions and Break Conditions

Five conditions carry no band until evidence arrives. A laboratory or federal program stating a position on hosted quantum access would activate P-1’s condition. A published materials or chemistry model trained on quantum-computed labels that beats its classical-only baseline on a named target would establish the productivity claim P-2 does not test. A demonstrated advantage on an industrial chemistry problem would convert P-2 from a disclosure test into a revenue test.

A room-temperature installation under S-3 would move weight toward distributed islands and weaken P-1’s base route. A FERC co-location order at nuclear hosts would decide whether Richland or a geothermal site hosts the first hub.

Three events would break the set. A reversal of the dequantization results, meaning a demonstrated quantum advantage on general machine learning, would move quantum onto the GPU side and void the paper’s central claim. A hyperscaler acquiring regional anchors below 50 megawatts would collapse the customer tier and the shared-hub route together. A DARPA finding before 2030 that no final-stage team reaches utility scale would push every window past the horizon without changing the anchor’s classical case.

IX. Strategic Insights by Audience

One design choice runs under every audience’s decision: build the anchor on classical economics and reserve the island, or build the island first and hope the tenant follows. Policymakers meet the choice in protected-load definitions. Executives meet it in hub terms, counsel in tenancy clauses and investors in capacity credit.

Strategic Posture by Audience

🏛️ Policymakers. Quantum adds one continuity objection and no energy relief. Define protected load by function and recovery time rather than by technology label. A federal research program’s multi-site procurement is the lever that forms a hub, so a state that wants the hub should align its laboratory’s hosting authority with it (P-1, S-2).

A hub is a procurement outcome before it is an energy outcome, and policymakers should treat it as one. A laboratory that buys quantum access for several sites forms a hub by contract, and no state incentive reaches that decision. The state’s lever is the hosting authority of its own institutions, and the observable is a laboratory or university hosting position stated in writing.

💼 Executives. Hosts should join a shared hub before installing an island and isolate refrigeration and control on a protected circuit before the first tenant installs. Hosts should sell the island on per-candidate research output with a named laboratory partner (P-1, P-2).

Installation speed is the fact executives should weigh first. An island installs in two quarters and a hub needs about six to serve three sites, so the island wins the race whenever a vendor offers placement terms, and the host that accepts them sets a template the region follows. Hosts that negotiate hub access first convert the asymmetry into a waiting option they control.

⚖️ Counsel. Training labels and the models trained on them are the island’s highest-value output, and research access is the other revenue channel. Allocate ownership, publication rights and export-control treatment of quantum-produced data in the tenancy before the first label exists. Write the ride-through duty and the protected reserve into the hosting agreement (P-2).

Counsel are allocating a research output rather than a capacity, so the drafting unit is the label and the model, not the rack. Export-control treatment of quantum-produced data remains undetermined, and the agreement should assign who bears a change in it. A tenancy drafted from a colocation template assigns the rack and leaves the output unowned.

📊 Investors. Quantum readiness has option value and zero near-term capacity value. Underwrite the island at ten to twenty percent utilization until a hub serves three sites, and value its output per validated candidate once a trained model exists (P-1, P-2).

Investors should expect the island’s revenue per validated candidate, not per megawatt, which changes the underwriting unit with the asset. A case that applies capacity terms to a research instrument misstates both utilization and revenue timing. The three-site hub contract is the first observable that converts option value into capacity value.

Where Audience Interests Collide

Hosts and vendors want different placements. A vendor earns on each installed island and a host earns on utilization, so the vendor’s placement offer and the host’s hub interest point in opposite directions until a hub serves enough sites to cover fixed cost. P-1 resolves in the hub’s favor only where a multi-site buyer arrives before the vendor’s offer closes.

Laboratories and hosts split on ownership. A laboratory wants publication rights and the models trained on its labels, and a host wants a revenue share in every later discovery. The tenancy decides the split, and P-2 reads the first disclosure as the evidence of which party won it.

Counties and hosts size continuity differently. A county wants continuity conditions sized to the site, and a host wants them sized to the island’s measured load. The functional protected-load test resolves the scope, and S-3 shows whether the processor needed the condition at all.

Takeaway. Each audience’s best move follows from treating the island as a research tenant, and the collisions resolve on dates the Simulation Predictions already carry.

X. Risk Mitigation by Simulation Prediction

Every entry below pairs a Simulation Prediction with the exposure it creates for one audience, the actions that audience can take alone and the exposure that survives. Severity and probability run on separate axes throughout.

Mitigating Actions by Time Gate

• Now to 30 days: strip unmeasured quantum efficiency claims from every pending filing and investor document.

• 30 to 90 days: negotiate hub access terms with at least one vendor before any island quote is accepted.

• 3 to 12 months: write the tenancy with a named laboratory partner and a named first model, and write label ownership into it.

• Beyond 12 months: keep the anchor’s protected circuit sized for an island so one can be added without redesign if the hub route fails.

Exposure, Actions and Residual for Each Simulation Prediction

The actions are analytic options for each audience and not legal, investment or fiduciary advice to any party.

Entry 1 · 💼 Host executives · P-1 (48–60% given the buyer).

• Exposure: a host that installs a local island before joining a shared hub carries $3.5 million to $4.5 million a year of fixed cost at 10 to 20 percent utilization, for the quarters until a tenant arrives. Severity: high. Probability: moderate.

• Mitigating actions: the commercial lead negotiates hub access terms with at least one vendor before any island quote is accepted, by the first 2027 hosting cycle. The engineering lead designs the protected circuit and ride-through so an island can be added later without redesign, before the first operator agreement. The finance lead models the island on per-result and access revenue only, with zero capacity credit, before any capital commitment.

• Residual exposure: a vendor can offer placement terms that make a local island rational for one host and irrational for the region, and the host that accepts it sets the template others follow.

Entry 2 · 📊 Investors · P-1 (48–60% given the buyer).

• Exposure: an underwriting case that credits a local island with hub-level utilization overstates annual margin by the island’s fixed cost, $3.5 million to $4.5 million at an anchor. Severity: moderate. Probability: moderate.

• Mitigating actions: the credit committee underwrites quantum at 10 to 20 percent utilization until a hub contract with at least three remote sites exists, and tests the case against the six-site break-even in Section VI. The diligence lead requires the hosting agreement’s access terms before closing.

• Residual exposure: utilization at the first hub may still fall short of the level that covers fixed cost.

Entry 3 · 🏛️ Policymakers · P-1 (48–60% given the buyer).

• Exposure: a commission that grants protected-load status to quantum equipment by label invites every later technology to claim it, and a county that conditions the first island on campus-scale continuity sets a cost no shared hub can meet. Severity: moderate. Probability: moderate.

• Mitigating actions: commission staff publish a functional test for protected load keyed to recovery time before the 2027 sessions. County planning staff size continuity conditions to the island’s measured load rather than to the site.

• Residual exposure: a statute naming quantum equipment can override functional tests.

Entry 4 · 💼 Research tenants and host executives · P-2 (44–56%).

• Exposure: a host that sells the island on energy or efficiency claims rather than on training-data output has no revenue line when the first disclosure shows no relief, and loses the research tenant that would have funded it. Severity: high. Probability: moderate.

• Mitigating actions: the commercial lead writes the tenancy as per-candidate or per-label output with a named laboratory partner, before the island is specified. The research lead names the first materials or chemistry model the island’s labels will train, in the tenancy.

• Residual exposure: the first model trained on quantum labels may underperform the classical baseline on the chosen target.

Entry 5 · ⚖️ Counsel · P-2 (44–56%).

• Exposure: a tenancy silent on ownership of quantum-produced labels and the models trained on them leaves both parties exposed to a dispute over the island’s highest-value output. Severity: moderate. Probability: moderate.

• Mitigating actions: transaction counsel allocates label ownership, model ownership and publication rights in the tenancy before any data is produced. Regulatory counsel documents the client’s assessment of export-control treatment for quantum-produced data before the first cross-border transfer.

• Residual exposure: export-control treatment of quantum-produced data remains undetermined and can change by rule.

Takeaway. Building the anchor on classical economics protects every audience under both outcomes of P-1, because a hub still attaches if islands spread and the anchor still earns if quantum stalls.

XI. Implications Run From Hub Formation to Label Ownership

Geothermal and hydro hosts capture the first research tenants and the first hub placements, because they alone firm the island at modular scale among clean sources (P-1). Vendors gain from hub placement where utilization covers their support cost, and lose from islands that sit idle. The Richland corridor holds the one place where a reactor and hydro coincide with a laboratory and a tenant under separate ownership. The federal co-location docket decides whether it hosts the first hub.

Training labels and the models trained on them are the island’s highest-value output through 2032 (P-2), and research access earns alongside them. The party that owns the labels captures the value of every later discovery, and the tenancy decides that party. Counsel’s drafting in Section X is therefore the distribution decision for the cluster’s quantum phase.

If P-1 misses, distributed islands spread before any hub serves three sites. Vendor placement incentives then dominate utilization, and the cluster model should expect higher fixed cost per host and slower hub formation.

If P-2 misses, the first island reports an energy metric first. The miss signals that hosts sold the island on efficiency claims the quantum overview expects to fail. The training-data loop then arrives later through research tenancies rather than through host disclosures.

Takeaway. A miss at either Primary Simulation Prediction delays the quantum phase without changing the anchor’s classical case.

What to Watch: The First Clean-Hosted Installation and the Buyer Who Contracts for It

The dominant fork is whether a multi-site research buyer contracts before a second island installs (P-1).

• Through 2027. DARPA’s final-stage results and any laboratory or federal statement on hosted quantum access; the latter activates P-1’s condition.

• December 31, 2029. S-3 resolves against colocation installation specifications.

• December 31, 2030. S-1 resolves against published benchmarks; S-2 resolves against the first hub’s tenant announcement.

• December 31, 2031. P-1 resolves against vendor and host releases, given the buyer.

• December 31, 2032. P-2 resolves against host, vendor and laboratory publications.

Verification sources by Simulation Prediction.

• Resolves P-1: quantum hosting announcements at clean energy sites and hub access contracts, reviewed quarterly in vendor and host releases.

• Resolves P-2: workload disclosures from clean-hosted islands and their tenants, reviewed quarterly.

• Resolves S-1: dispatch and scheduling benchmarks, reviewed at each vendor or peer-reviewed publication.

• Resolves S-2: tenant announcements at the first hub, reviewed quarterly.

• Resolves S-3: colocation installation specifications, reviewed quarterly in operator publications.

• Moves watch conditions: industrial chemistry advantage results, FERC co-location orders and DARPA final-stage results, reviewed as they appear.

Conclusion: Quantum Joins as a Tenant, Becomes a Data Source and Does Not Run the Model Within the Horizon

Quantum computing does not accelerate inference or training within the horizon, and the cluster should not claim that it will. It joins a regional clean compute anchor as a research tenant, becomes a source of training labels that let scientific models reason about correlated matter, and earns only as a shared hub or under a public tenancy. The island arrives in 2031 to 2032 and the reactor follows in the mid-2030s, and the Richland corridor is where they coincide.

Stakeholders should act now on what they control: hub access terms before any island quote, a tenancy with a named laboratory and a named first model, and label ownership written before the first label exists. Stakeholders can wait to learn whether DARPA’s final-stage teams reach utility scale and whether a room-temperature processor makes the island an ordinary rack.

The decisive signal is the first quantum hosting arrangement on a clean energy site. If it serves three remote sites before a second host builds its own island, P-1 holds and the shared-hub route governs the quantum phase. If a second island installs first, P-1 misses and the model expects higher fixed cost and slower hub formation while the anchor’s classical case stands.

Working With MindCast

Public analysis cannot tell a specific host whether its anchor should host a hub, join one or wait. The answer depends on the host’s firm capacity, its laboratory partner’s budget, its county’s continuity posture and the vendor terms on offer. Site inputs stay confidential.

MindCast builds Cognitive Digital Twins of the host, its candidate vendor, its research tenant and its county. The engagement ingests the organization’s own decision history and reruns the hub-versus-island contest and the tenancy terms against its actual options.

Commissioned scope adds complete probability distributions, P10/P50/P90 ranges, organization-specific calibration and complete scenario trees. The scope also adds utilization modeling for a hub serving the host’s named remote sites and red-team tests of vendor placement offers.

Three engagements map to this installment’s readers. A Hub-or-Island Diagnostic for a host or a quantum cohort member: one anchor, one vendor and one tenancy decision (P-1, S-2).

A Tenancy Terms Review for counsel tests label ownership and publication rights together with export-control treatment and continuity duties against vendor and laboratory responses (P-2, S-3). A Protected-Load Brief for a committee: the functional test run against county and commission Cognitive Digital Twins (P-1). Contact mcai@mindcast-ai.com or visit www.mindcast-ai-simulation.com to arrange one.

Appendix A. MindCast Works

Companions

Regional Clean Compute Anchors, How Much Compute Each Clean Energy Source Yields and Where Washington Builds the First One (Installment One). Sets the anchor configuration, the compute yield per source and the allocation sequence. Supplies the firm-clean host on which the island sits.

How Quantum Improves AI, Five Channels Ranked by Date and Value, and the One That Pays for a Clean Compute Anchor (Installment Three). Ranks the five channels and states the ownership mechanism for quantum-produced labels. Supplies the remote-delivery finding that resolves this installment’s hub-versus-island contest in the hub’s favor.

How Quantum Technology Changes the AI Data Center Energy, Compute and Environmental Calculus (October 2026). The quantum series overview. Supplies the quantum island, the finding that added output absorbs energy savings and the base route of delayed usefulness after 2030 that this installment’s phasing follows.

Funded Solutions That Can Help Win Quantum-Coupled AI Data Centers Approvable (October 2026). Tests 31 measures across three project types. Supplies the quantum cost delta, the hub approval share and the separation of quantum-inspired software from quantum hardware.

Where the Clean Energy Host Advantage Holds and Where It Breaks in AI Data Centers (October 2026). Cover letter on host sites. Supplies the hyperscaler hosting forecast behind the customer tier and the finding that county review holds at energy sites.

AI Data Centers and Clean Energy: Which Projects Reach Operation and Where Clean Technology Companies Can Win(October 2026). The magazine companion. Supplies the segment guide that this installment extends to quantum vendors and research tenants.

The series umbrella and the market

AI Data Center Authorization + Clean Energy, the Mid-Scale Squeeze and the Specialized Compute Landscape(September 2026). The series umbrella. Supplies the Mid-Scale Squeeze that defines the island’s customer tier.

AI Datacenter Edge Computing: Ship the Workload Not the Power (September 2025). The founding principle of moving compute toward energy. Supplies the siting logic that places the hub on the firm clean host rather than in the metro.

The Grid-Anchored Clean Power Bargain: Powering AI Data Centers Through 2040 (September 2026). Establishes the firmness inversion. Supplies the reason the island cannot sit on a curtailable campus.

The Clean Compute Match: Which Clean Power Pairs With Which AI Compute, and What Stays on Fossil (September 2026). Sorts compute and clean generation into matched pairs. Supplies the pairing that puts the island’s continuous load on firm supply.

The approval layer

Burden-Based Authorization for AI Data Centers, Modular Clean Compute, and the Backlash Gap Between Grid Tariffs and State Law (September 2026). Defines the four burden measures. Supplies the measured-burden posture the island must carry at the county.

AI Data Center Veto Points Show Where Projects Get Stopped and How Authorization Risk Moves (September 2026). Maps ten veto nodes. Supplies the continuity node the island adds.

The Lowest-Exposure AI Data Center Approval Route for Developers, Hyperscalers and Investors (September 2026). Ties capital release to durable approvals. Supplies the staging rule behind zero capacity credit for unmeasured quantum claims.

The Data Center Authorization Market: A 50-State Regulatory Atlas (August 2026). State-by-state evidence on approval terms. Supplies the protected-load and critical-load provisions the policymaker entries address.

The Model Data Center Authorization Code (July 2026). Compares any state’s provisions by function. Supplies the common language for the functional protected-load test.

The assembly

Modular AI Data Center Specialists and the Integrated Offerings That Turn Small Sites Into Authorized Operating Capacity (September 2026). Identifies power-asset hosts as the party that carries small sites to operation. Supplies the host that captures the quantum tenant.

AI Data Center Taxonomy for Decision-Making and Clean Energy, Four Categories That Decide Whether a Site Operates(September 2026). Replaces facility labels with four decision categories. Supplies the “can prove” category through which the island’s measured burden enters the approval decision.

Appendix B. External Sources

• DARPA. “Four More Teams Enter Quantum Benchmarking Initiative Final Stage.” October 7, 2026. The six final-stage teams and the 2033 utility-scale question.

• Equinix and Diraq. “Diraq and Equinix to Deploy Australia’s First Quantum Computer Inside a Data Centre.”September 1, 2026. Sub-20-kilowatt draw with self-contained cooling.

• IonQ. “IonQ to Advance Quantum Supercomputing by Bringing First QPU to NVIDIA Accelerated Quantum Research Center.” September 23, 2026. Superion 256 and NVIDIA GB200 NVL72 coupling through NVQLink.

• Google Quantum AI. “Quantum Error Correction Below the Surface Code Threshold.” Nature, December 2024. Logical error falling with code size.

• IBM. “IBM Sets the Course to Build World’s First Large-Scale, Fault-Tolerant Quantum Computer.” June 2025. The qLDPC architecture and the target of about 200 logical qubits by 2029.

• Open Compute Project. “Integrating Quantum Processing Units into Data Center Infrastructure.” White paper, 2025. Modality classes and facility requirements for quantum deployment.

• HPCwire. “Colocating Quantum with Classical Computers: Building New Hybrid Data Centers.” July 29, 2024. Vibration, cryogen and maintenance requirements.

• Data Center Dynamics. “When Data Centers Meet Quantum Computers.” Oxford Quantum Circuits deployments and the data-residency driver.

• Gidney, C. “How to Factor 2048-Bit RSA Integers With Less Than a Million Noisy Qubits.” 2025. The cryptographic timeline in Tier Three.

• Tang, E. “A Quantum-Inspired Classical Algorithm for Recommendation Systems.” 2018. The dequantization line of work.

• Energy Northwest and X-energy. Cascade Advanced Energy Facility near Richland in Washington State. Up to twelve Xe-100 modules with Amazon investment announced October 2024.

Companion works: What Quantum Computing Contributes to Clean Compute Clusters · How Quantum Improves AI · Where the Clean Energy Host Advantage Holds and Where It Breaks in AI Data Centers · AI Data Center Authorization + Clean Energy, the Mid-Scale Squeeze and the Specialized Compute Landscape

Why This Installment Matters to Each Audience

Each audience faces one decision the installment changes, and the cost of deciding without it differs by audience.

🏛️ Policymakers. State committees and county boards inherit the anchor question in 2027 whether or not they take it up, because the first host filing forces a classification. The installment supplies the compute-yield table that tells a committee how much operating capacity each definition releases, and it names the four instruments that move the outcome. A nuclear-only definition leaves hydro and geothermal anchors in negotiation, and the simulation treats that as the likeliest 2027 result.

💼 Executives. Developers and hosts gain a sorting rule they can apply before any site visit: firm fraction decides whether a site sells inference or interruptible compute, and no later negotiation changes it. The installment states the compute each source yields per 100 megawatts and the cluster size an anchor can reach, which sets the revenue envelope before the first agreement. Executives who skip the sort spend a permitting cycle on a site that cannot host the tier they planned to sell.

⚖️ Counsel. Hosting agreements in the current market are silent on import rights at module load and on allocation of approval delay. The installment identifies both clauses and the federal co-location question at nuclear hosts, and it states which Simulation Prediction each clause protects against. Counsel who draft from campus templates import campus burdens into an accessory-use site.

📊 Investors. The anchor’s revenue rests on two products with different firmness and different counterparties, and the staging gates differ by source. The installment supplies the gates: the executed allocation agreement at hydro sites, the construction permit at reactor sites and the research tenancy only once signed. Investors who underwrite the anchor on campus comparables misstate both scale and schedule.

Takeaway. The installment gives each audience the one classification its decision turns on, and Sections VII and VIII state what to conclude and what to do with it.

I. Clean Energy Source Sets the Compute, and Only Three Sources Firm an Anchor

Clean energy source decides how much compute a modular site can host, and only three clean sources firm enough of it at modular scale to anchor a regional cluster without multi-day storage or grid firming. Geothermal, hydro and the first small modular reactors yield continuous compute at a scale a county can approve as an accessory use. Solar and wind yield more compute per dollar but only for work that can pause. Washington holds all three anchor sources within three milliseconds of two metros and a national laboratory.

Power allocation is the binding economic constraint there even where a county acts first, because a county approval without import rights produces no operating capacity.

Thesis. A regional clean compute anchor is a firm-clean site of 8 to 20 megawatts that hosts enterprise inference for a nearby metro and pairs with solar and wind modules that carry movable work. The anchor’s economics clear on two products: interruptible clean compute at the variable modules and firm regional inference at the anchor.

The anchor reserves under one megawatt for a protected quantum island, and Installment Two states what that island contributes. A small modular reactor firms the anchor at three times the hydro scale. The Richland reactor plans construction by 2030 and operation in the mid-2030s, so a reactor-firmed anchor arrives after the island has already landed on hydro or geothermal power.

Why now. Energy Northwest and X-energy plan up to twelve 80-megawatt reactor modules beside Columbia Generating Station near Richland. Construction is planned by 2030 with operation targeted in the 2030s, with Amazon investment announced in October 2024. The paper models operation in the mid-2030s, later than the early-2030s target in X-energy’s public filings.

Grant County’s public utility district capped data center load growth in April 2025 on transmission grounds, with relief lines due in 2027 and 2029. Washington’s clean technology committees, the state’s clean energy developers and the counsel who advise them each decide before the 2027 session writes the next terms. On October 9, 2026 the Federal Energy Regulatory Commission (FERC) rejected PJM Interconnection’s curtail-first plan for large loads and left retail curtailment with the states.

Energy Vault broke ground on an 8-megawatt modular campus in Snyder, Texas on July 27, 2026. Texas halted data center permits eight weeks later. Washington’s 2026 tax change, Senate Bill 6231, removed the incentive exemption for refurbished facilities while the 2027 session prepares to write the next terms.

What distinguishes the MindCast approach. Engineering firms size the power and counsel reads the rules. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through MindCast AI Proprietary Cognitive Digital Twin Foresight Simulations (MP CDT FS). Each Cognitive Digital Twin (CDT) models how a named institution decides as evidence and rules change, and Section V names the twelve this installment runs.

Behavioral Economics supplies the decision rules: counties react to the salient label while utilities apply thresholds regardless of salience. Game Theory supplies the payoff structure: an anchor survives only when the host beats its outside option and the allocation authority recovers its cost. Predictive simulations emerge from the combination.

What the paper does. Section II converts each clean source into compute and sets nuclear’s three doors. Section III sets the anchor configuration, its workload envelope, its size against a campus and the fixed-versus-reconfigurable contest the simulation leaves open. Section IV states the three conditions that place anchors, lists the six regions that meet them and works the Washington case.

Section V states the simulation architecture. Section VI carries the MindCast Foresight Simulation Predictions with their bands. Section VII states the strategic insights for each audience, and Section VIII carries the Risk Mitigation entries.

Simulation Prediction highlights. Six Simulation Predictions in Section VI.

• P-1. A Washington public utility district or utility executes a service agreement for a modular compute anchor at an existing hydro or nuclear interconnection, through December 31, 2028. 45–57%.

• P-2. The first quantum-equipped project (the island Installment Two describes) to receive a county decision in any county of the six candidate regions is a firm-clean modular anchor under 20 megawatts rather than a campus addition, through December 31, 2029. 42–54%.

• S-3. A small modular reactor site announces a modular compute anchor with a named compute operator, through December 31, 2030. 40–52%.

Stakeholder readout.

• 🏛️ Washington legislators on the clean technology and nuclear committees. Three definitions in the 2027 session decide whether the anchor tier forms in Washington or in geothermal states: added load at an existing generation interconnection, accessory computing use at an energy facility, and the tax classification of compute added inside an existing facility. Treat the Richland reactor site as the first nuclear unit sized to those definitions (P-1, S-3, S-4).

• 💼 Clean energy developers and hosts. Geothermal, hydro and reactor hosts hold the anchor role and sell regional inference. Solar and wind hosts hold the module role and sell interruptible compute. Classify every site by firm fraction before any county or lender conversation (P-1, P-2).

• ⚖️ Counsel to clean energy hosts. The anchor adds one allocation question at hydro hosts and one federal co-location question at nuclear hosts. Write import rights and delay allocation into the hosting agreement before the first operator signs, together with phase thresholds and the tax classification (P-1, S-3).

• 📊 Investors. Underwrite the anchor on regional inference and interruptible compute. Stage capital to the allocation agreement at hydro sites and to the construction permit at reactor sites, with reactor operation in the mid-2030s (P-1, S-3).

II. Compute Yield Per Clean Energy Source Decides Which Sites Can Anchor a Cluster

Nameplate megawatts are not compute megawatts. Three conversions sit between them, and they sort clean sources into anchors and modules before any approval question arrives. The section ends with nuclear, which reaches the anchor model by three routes and fits the anchor through one.

Three Conversions Separate Nameplate From Compute

Capacity factor turns nameplate into average energy. A 100-megawatt solar plant averages about 25 megawatts across a year and a wind plant 35 to 40. A dispatchable hydro plant averages 40 to 50 and a geothermal or nuclear plant 85 to 92.

Firmness then turns average energy into load that can run every hour. Storage raises a variable plant’s firm share, but a four-hour battery cannot carry a winter night, so solar and wind firm only a small fraction of their average output. Storage round-trip losses take about five percent off the firm megawatts that pass through a battery.

Power usage effectiveness turns facility power into information technology (IT) load. Air-cooled modular sites run near 1.2, so a 20-megawatt facility delivers about 16.5 megawatts to servers. A liquid-cooled rack of 72 current-generation accelerators draws about 120 kilowatts, which puts 550 to 600 graphics processing units (GPUs) on each megawatt of IT load.

The chain therefore runs in one direction with one denominator at each step. Nameplate times capacity factor gives average megawatts. Average times firm share gives firm megawatts and firm less storage loss gives facility megawatts. Facility divided by 1.2 gives IT megawatts and IT times 550 to 600 gives GPUs.

A 20-megawatt facility anchor runs 9,000 to 10,000 GPUs, and an 8-megawatt facility anchor runs 3,500 to 4,000.

Takeaway. The source sets the firm fraction and the firm fraction sets the continuous compute. Continuous compute is the number a county, a utility and a lender each read.

Compute Yield Per 100 Megawatts of Nameplate

The table separates four quantities that a megawatt figure conflates. Average megawatts is annual energy. Firm megawatts is the load the source can hold every hour at a stated reliability target. For hydro the firm figure cannot exceed average energy across a year however dispatch moves it between hours.

For solar and wind the firm figure does not follow from capacity factor and four-hour storage alone. Storage duration and hourly generation set it together with the reliability target and any backup. Installed GPUs is hardware, and usable GPU-hours at 85 percent availability is the closest the table comes to work.

Completed useful work depends on the workload mix and belongs to a site-specific engagement. Every figure is illustrative until checked against a site’s hourly generation data.

Geothermal and hydro firm a 20-megawatt anchor from a modular-scale plant: about 25 megawatts of geothermal nameplate or 45 to 60 megawatts of hydro. Solar would need over 300 megawatts of nameplate and multi-day storage to firm the same 20 megawatts, which no host will build for a module. Solar and wind therefore supply the movable majority of a cluster’s compute at the lowest power cost per GPU-hour, and they cannot host the inference tier. The table compares power cost per GPU-hour and not total cost per completed job, which also depends on checkpoint overhead and idle time.

Takeaway. Only geothermal, hydro and nuclear build the anchor, and only the first two build it at the scale a county treats as accessory use.

Nuclear Enters Through Three Doors and Fits the Anchor Through One

Nuclear reaches the anchor model by three routes, which the paper calls doors: an existing plant hosting modules, a small modular reactor sized to the anchor and a campus-scale plant.

Existing plants host modules the way Three Mile Island and Palisades host hyperscaler capacity. The module inherits the plant’s interconnection, land and firm output. The veto point is federal, not local: behind-the-meter co-location at a nuclear plant opened the PJM and FERC dispute over who pays for transmission the plant no longer uses for export.

MindCast’s host-advantage simulation Where the Clean Energy Host Advantage Holds and Where It Breaks in AI Data Centers found that the county advantage holds at energy sites and that the federal layer binds first at a nuclear host. The finding matters here because it places the federal question ahead of every local one at reactor sites.

A small modular reactor (SMR) is the first nuclear unit sized to the anchor. One Xe-100 or NuScale module runs 77 to 80 megawatts, which could firm a 60 to 70 megawatt anchor as an allocation scenario, three times the hydro or geothermal anchor.

Energy Northwest’s project near Columbia Generating Station plans up to twelve modules and 960 megawatts, with construction by 2030, and it remains in pre-application with the Nuclear Regulatory Commission. The paper models operation in the mid-2030s as its schedule, and every reactor-route date below rests on that modeled schedule rather than on the project’s own target. No compute allocation from it has been proposed. A reactor-firmed anchor raises compute yield from 9,000 GPUs to 30,000 or more at a single module and removes the water-year risk that caps the hydro band.

Campus-scale nuclear builds campuses rather than anchors. Gigawatt nuclear contracts firm frontier training and consumer inference at the scale that passes every approval gate by self-funding. Burden-based approval does not apply to that tier, and the paper does not argue for it. Nuclear firm power costs more per megawatt-hour than hydro or geothermal and less than any gas-firmed renewable.

Takeaway. Nuclear belongs in the anchor model through the small modular reactor door, and Richland is where that door opens first.

III. The Regional Anchor Configuration and the Workloads It Can Carry

Three constraints select the configuration, and the workload envelope and the cluster’s size follow from them. The section closes by naming who rents the cluster and why hyperscalers do not build it.

Three Constraints Select the Configuration

Firmness comes first. Anchored inference cannot tolerate intermittency, so the host must be firm clean. The 20-megawatt line comes second. Above it a site enters large-load study at most utilities, whose thresholds run from 20 to 50 megawatts, so each site stays below the line and scale comes from a portfolio of sites.

The county’s accessory-use reading is a separate determination governed by local code, and no size line secures it. The line is a conservative design choice against the utility threshold rather than a legal protection at the county.

Coupling radius comes third. Work that needs a shared fabric stays inside one building, and work that tolerates milliseconds moves between sites. The configuration the three constraints select has three parts. One firm-clean anchor of 8 to 20 megawatts holds the enterprise inference tier, two to five variable modules of 2 to 8 megawatts sit at solar and wind sites with storage and carry movable work, and one scheduler assigns work across every site.

The whole set operates as one Authorized Compute Portfolio of 20 to 50 megawatts, the series umbrella’s term for several small sites approved and scheduled as one operating unit, and never as one campus. The anchor reserves a protected circuit for a quantum island under one megawatt. Installment Two is What Quantum Computing Contributes to Clean Compute Clusters, Tenant First, Training Data Second and Not the Processor Through 2032. The companion states what the island contributes and why the anchor’s economics never depend on it.

Takeaway. The anchor is the smallest unit that hosts firm regional inference below most utility study thresholds, and the county reads it on measured burden rather than on size.

The Workload Envelope

A 9,000-GPU cluster is small for training and large for inference. The paper assumes that scale serves a regional enterprise customer base at production quality, and a workload-specific service benchmark would confirm it. The configuration therefore competes with metro colocation on firmness and burden rather than with hyperscale campuses on scale.

Takeaway. The cluster should claim regional inference and movable training, and it should not claim consumer or frontier work outside the latency radius.

One Scheduler, Many Sites, Job-Level Coupling in Milliseconds

Light in fiber covers about 200 kilometers per millisecond. Modules 50 kilometers apart exchange a message in roughly half a millisecond round trip. A GPU fabric inside a rack runs at microseconds and terabytes per second per accelerator, and a synchronized training run needs that fabric.

Modules can therefore share jobs, route inference, migrate checkpointed work and pool a scheduler. Modules cannot form one training fabric across sites. The scheduler works on three timescales: within the hour on which jobs run or pause, daily on how much power goes to inference against batch work, and quarterly on which physical module to add or retire.

Takeaway. The coupling claim that holds is one scheduler across many sites with job-level exchange in milliseconds, which covers every workload the envelope admits.

Fixed or Reconfigurable Cluster Is a Contest the Simulation Leaves Open

Flexibility carries two economic values, and MindCast’s series has developed only the first. Workload mobility lets movable work follow variable power and reduces dependence on scarce firm supply. Hardware reconfiguration lets a cluster add or retire GPU and power modules as supply and accelerator generations change. Reconfiguration preserves capital value when demand shifts.

The contest is fixed modules with software scheduling against a cluster that reconfigures its hardware. The simulation holds electricity supply and customer demand and available capital constant. It varies orchestration and hardware composition along with quantum access and switching cost. Operators lean toward fixed modules by commissioning cost, and enterprise buyers lean toward reconfiguration only where service guarantees survive it.

If reconfiguration wins, the unit of investment changes from a facility to a managed portfolio and the anchor’s value exceeds its power pairing. If fixed modules win, the cluster’s value reduces to its power pairing and the compute-yield table is the whole economic case. The simulation could not assign a value, because no operator has disclosed workload traces or module exchange costs. The question resolves on matched throughput and total system cost across the two configurations on an operator’s own traces.

Takeaway. Reconfiguration is the one variable that decides whether the cluster is worth more than its power, and an operator’s own traces decide it.

Cluster Size Against an Ordinary Data Center Campus

One 20-megawatt anchor runs about 130 liquid-cooled racks and 9,000 to 10,000 GPUs. A 300 to 500 megawatt campus runs 2,000 to 3,500 racks and 150,000 to 250,000 GPUs. One anchor is therefore one twenty-fifth to one sixtieth of a campus. A full portfolio of 20 to 50 megawatts installs 25,000 to 40,000 GPUs and delivers roughly 15,000 to 20,000 GPU-equivalents of continuous throughput, one sixth to one tenth of a campus.

Water at an air-cooled modular site runs near zero. Energy Vault’s 33 planned modules at Snyder, Texas use about 297 gallons a day for the whole site.

Takeaway. The cluster trades scale for firmness, burden and coupling, and the trade is visible in every unit a county or lender reads.

Who Rents the Cluster and Why Hyperscalers Do Not Build It

Hyperscalers pass every approval gate by scale, and their workloads are frontier training and consumer inference at a size no 20-megawatt anchor serves. MindCast’s host-advantage simulation holds that none of the four largest operators hosts compute at a third-party energy site below 50 megawatts through 2027.

The customer is the tier the Mid-Scale Squeeze strands: buyers too large for cloud spot capacity and too small to command a campus, as MindCast’s series umbrella AI Data Center Authorization + Clean Energy, the Mid-Scale Squeeze and the Specialized Compute Landscape defines them. Enterprises with residency constraints and neoclouds form one group. Sovereign and research buyers and clean energy companies monetizing their own power form the other.

Renting clears on two products. Interruptible clean compute at solar and wind modules undercuts cloud spot rates for batch and fine-tuning because power is the lowest-cost input in the model’s stack. Firm regional inference at the anchor competes with metro colocation on burden, not on cost. Nobody rents a modular site for uptime it cannot promise, so the anchor sells only the firmness its source can deliver.

Takeaway. The cluster differs from a campus in its unit of approval and its power pairing rather than in its hardware, and its customer is the tier campuses leave behind.

IV. Three Conditions Place Regional Anchors in Six Regions, and Washington Holds the Strongest Case

Regional anchors form where three conditions coincide, and the map of those conditions already exists in the Department of Energy’s research footprint. The section closes with Washington, where allocation decides before the county does.

The Three-Condition Rule

A regional anchor forms where a firm clean source sits within about three milliseconds of fiber, roughly 300 kilometers, of a metro of two million people or more. The same radius must hold a research institution that can anchor a specialized tenant. The local utility or public power authority must still hold allocatable firm capacity. Latency sets the radius, firmness sets the source, and allocation decides whether the anchor gets built.

Variable clean regions supply movable modules to a portfolio and cannot host an anchor. Texas therefore enters the model as a module region rather than an anchor region.

Takeaway. Three conditions rather than fifty state rules decide where anchors form.

Six Regions Meet the Rule

• Pacific Northwest: Columbia River hydro and the Richland reactor project serve Seattle and Portland. Pacific Northwest National Laboratory and the University of Washington anchor the tenant.

• Northern California and Nevada: Geysers and Great Basin geothermal serve the Bay Area and Reno. Lawrence Berkeley National Laboratory and Stanford anchor the tenant.

• Great Lakes: the Illinois nuclear fleet and Michigan restarts serve Chicago and Detroit. Argonne and Fermilab anchor the tenant with the Chicago Quantum Exchange.

• Tennessee Valley: Tennessee Valley Authority hydro and nuclear serve Nashville and Atlanta. Oak Ridge National Laboratory anchors the tenant.

• Mountain West: Utah and Idaho geothermal serve Salt Lake City. Idaho National Laboratory and the Utah and Colorado programs anchor the tenant.

• Mid-Atlantic: Pennsylvania and Virginia nuclear serve Philadelphia and Washington. The Maryland institutes anchor the tenant. Metro saturation and Pennsylvania’s township veto make it the hardest of the six.

The paper ranks none of the six, because a ranking needs one definition each of metro proximity, research demand and allocatable power.

Five of the six regions sit on a national laboratory or its university partners. The research tenant and the firm clean source were placed decades ago by the same logic: low-cost firm power drew the laboratories. The Great Lakes and Mid-Atlantic entries work as modules added at existing nuclear plants rather than as new 20-megawatt anchors.

Takeaway. The regional map and the national laboratory map are the same map.

Washington: Allocation Binds Even When the County Acts First

Central Washington hydro sits 150 to 250 kilometers from Seattle and about the same from Portland. Fiber round trip over that distance is 1.5 to 2.5 milliseconds. A consumer response from a large language model budgets hundreds of milliseconds, so a Columbia River anchor sits inside the consumer latency radius for Seattle, Portland and Vancouver. The three metros form a combined market of eight million people, and hyperscalers have run inference in Quincy on this geography for fifteen years.

The binding constraint is hydro allocation rather than latency. Grant County’s public utility district capped data center load growth in April 2025 and named transmission as the limit instead of supply. Relief lines are due in 2027 and 2029.

Chelan and Douglas power is largely contracted and the Bonneville Power Administration’s new-load queue is long. Washington’s 2026 tax change removed the exemption for refurbished facilities while new construction keeps it.

Richland adds the reactor door. Columbia Generating Station and the planned Xe-100 modules sit on the Hanford site north of the city, Columbia River hydro runs past both, and Pacific Northwest National Laboratory’s campus sits in north Richland under separate federal ownership. The assets are related by geography and not by ownership or service arrangement, and the simulation treats them as four actors in one corridor. No other corridor in the country holds an operating reactor, a planned small modular reactor, dispatchable hydro and a national laboratory within one allocation authority’s reach.

Richland also carries two doors in one corridor, and which opens first is the live question. The hydro door opens if a hydro-firmed anchor beside the laboratory operates before the reactor. The reactor door opens if the anchor waits for the Xe-100 modules. FERC’s co-location order and the Nuclear Regulatory Commission’s schedule decide between them, and Energy Northwest leans toward the reactor by project identity while the laboratory leans toward hydro by schedule.

Takeaway. Richland holds both doors, and the first FERC co-location order decides which opens first.

The 2026 Incentive Change Needs a Classification Test Before It Counts Against Washington

Washington’s 2026 session ended the data center sales-tax exemption for qualifying refurbishment from July 1 while retaining the qualifying new-construction route. Whether an anchor that adds compute inside an existing energy facility falls under either route is a classification question the governing tax provisions decide, not a consequence that follows from the facility’s age. The anchor may qualify as new construction, fall under the ended refurbishment route, or qualify for neither.

A zoning definition of accessory use does not by itself restore a sales-tax exemption, because the two statutes classify on different terms. If the anchor falls outside both routes, the remedy is a targeted change to the tax provisions rather than a land-use definition. The legislature’s nuclear working group holds both drafting opportunities, and the simulation models the tax outcome as a classification test with three branches.

Takeaway. The incentive change is a live commercial distinction for Washington anchors, and its weight depends on a classification that must run before the tax treatment enters any strategy.

What the 2027 Session Can Decide

The legislature holds four instruments that move the Simulation Predictions in Section VI, and each can pass without the others. A definition of added computational load at an existing generation or storage interconnection moves allocation from case-by-case negotiation to a rule and releases P-1 early. A definition of accessory computing use at an energy facility keyed to firm import capacity and measured burden carries S-4 and gives counties a test to read. A classification fix in the data center tax provisions decides whether compute added inside an existing facility sits under the new-construction route or outside both routes.

A functional protected-load test keyed to recovery time rather than technology label prepares the continuity question that Installment Two’s island raises.

The simulation places a nuclear-only accessory definition as the most likely 2027 outcome, because the Richland development interest and the local backlash against data center proposals near Cle Elum in Kittitas County pull the working group in opposite directions and a nuclear-site bill splits the difference. A statewide definition keyed to burden across generation types reaches hydro and geothermal anchors as well, which is where the first Washington agreement is likely to land.

• 🏛️ Legislators. Draft the four instruments as one burden-keyed package rather than a nuclear-site bill, because the first anchor agreement in Washington is likelier at a hydro interconnection than at the reactor site (P-1, S-4).

• 💼 Clean energy developers. File burden exhibits and import requests before the session cutoff, so the working group drafts against a live project rather than a label.

• ⚖️ Counsel. Prepare the tax classification analysis for an existing-facility anchor before any bill text circulates, since the remedy differs by which route the anchor falls outside.

• 📊 Investors. Hold reactor-sited commitments to the construction permit and treat the session’s definitions as the first durable approval a hydro anchor can carry.

Takeaway. Four definitions in one session decide whether Washington’s anchor advantage becomes an agreement or a research-tenant advantage only. Washington is the strongest anchor geography in the country, and the first build turns on a public utility district allocation and a legislative definition. A county acting first changes the schedule without changing which constraint binds.

V. Twelve Cognitive Digital Twins Test Whether the Anchor Gets Built

The simulation answers one question. Which configuration of clean energy source and modular compute produces the most continuous compute a county can approve as accessory use, and which constraint binds when a host tries to build it? The horizon runs from October 2026 through 2030, with Washington as the named forum and the other five regions as comparison cases.

Four Events Since September Change the Forum, the Veto and the Tenant’s Date

FERC’s October 9 order moved continuity terms for any protected load from the wholesale level to state forums, which changes the forum for the first instrument any anchor needs. Texas’s September 21 permit halt showed that a state executive can stop energy-site modules as easily as campuses, which adds a state veto the county-centered analysis omits. Washington’s 2026 incentive change ended the refurbishment route to the data center exemption, which reaches anchors only if the governing classification places them there. DARPA’s October 7 final-stage selection tied the research tenant’s budget to a dated federal milestone, which dates the tenant’s arrival to 2033 rather than to vendor roadmaps.

The Actor System

Mechanism

Game theory supplies the payoff structure. An anchor survives only when the host beats its outside option, the allocation authority recovers its cost and the county sees burden it can measure. A reactor host pays the federal co-location question before any local question. A hydro host pays the allocation question first and the county question second.

Behavioral economics supplies the decision rules. Counties react to the salient label, so an air-cooled module at an existing energy site draws little review. Utilities and public utility districts apply thresholds regardless of salience, and legislators anchor on the data center label unless a burden definition arrives before the session. Predictive behavior emerges from the combination.

Scenario Routes

Among institutional outcomes, the simulation places allocation as the binding constraint at hydro hosts and the federal co-location docket at nuclear hosts, whichever authority acts first in time. County accessory-use treatment for anchors under 20 megawatts is the base route. Label review reaching energy sites sits second, and a statewide accessory-use statute sits as the tail.

Among configuration outcomes, the hydro anchor with variable modules is the base route through 2030 and geothermal regions move first where districts cap. The reactor-firmed anchor sits as the mid-2030s route. A hyperscaler absorbing every anchor site sits as the tail.

The fixed-versus-reconfigurable contest returned no value for lack of operator traces.

Takeaway. The simulation resolves which constraint binds and leaves reconfiguration open for lack of operator traces, and the Simulation Predictions follow from the first contest.

VI. MindCast Foresight Simulation Predictions

Six Simulation Predictions follow, two primary and four secondary. P marks a Primary Simulation Prediction and S a Secondary Simulation Prediction. Each names an actor and an outcome with a window, a band, a falsifier and a verification source. No Simulation Prediction in this installment has reached its resolution date, so the validated tally stands at zero hits and zero misses.

Simulation Predictions at a Glance

Competing Explanations

Three accounts compete before any result prints.

Allocation binds. Firm import capacity decides whether an anchor operates whichever authority acts first, and counties read measured filings at energy sites as accessory use. Falsifier: a county pause or denial stops an anchor that already holds import rights.

Label review reaches energy sites. Counties treat modules at generation sites as new data centers, and geothermal states without district caps build first. Falsifier: no county stops a module at an operating energy site through 2028.

Hyperscalers absorb the tier. A campus operator contracts the reactor output and anchor sites below 50 megawatts never form. Falsifier: a hyperscaler announces a third-party energy-site deal below 50 megawatts.

The simulation favors the first account at 50 to 60 percent of the weight, the second at 25 to 35 and the third at 10 to 15.

Primary Simulation Predictions

P-1 | A Washington Allocation Authority Executes the First Anchor Agreement (45–57%)

Forecast statement: a Washington public utility district or utility executes a service agreement for a modular compute installation at an existing hydro or nuclear interconnection, at a first phase under 20 megawatts, with or without a state definition of added load in force.

• Band: 45–57%.

• Window: through December 31, 2028.

• Trigger conditions: a host files for import service at an existing energy interconnection in Washington.

• Explicit falsifier: no such agreement executes, or the first executed agreement is at a site above 20 megawatts.

• Verification source: public utility district board minutes, utility filings and host releases.

P-2 | First County Decision in the Six Regions Lands on a Firm-Clean Anchor (42–54%)

Forecast statement: the first quantum-equipped project to receive a county land-use decision in any county of the six candidate regions in Section IV is a firm-clean modular anchor under 20 megawatts rather than a campus or hyperscale addition. Quantum equipment is the test case because it carries the continuity objection Installment Two describes, so the first such decision shows whether a county reads an anchor by burden or by label.

• Band: 42–54%.

• Window: through December 31, 2029.

• Trigger conditions: a county docket in a named region lists a project naming quantum equipment.

• Explicit falsifier: the first such decision concerns a campus, a hyperscale site or a project above 20 megawatts.

• Verification source: county minutes and ordinances in the eighteen counties listed below.

Observable populations. P-2 resolves against a fixed list of eighteen counties in the six regions.

Secondary Simulation Predictions

Observable population for S-1. S-1 identifies qualifying anchors nationally and not only inside the six regions. Host and operator releases and executed service agreements supply the population, and a qualifying anchor must meet all three conditions in its definition. A project outside the six regions that meets them counts against S-1.

Watch Conditions and Break Conditions

Two conditions carry no band until evidence arrives. A FERC order resolving co-location cost allocation at nuclear plants would move weight toward the reactor route early. A Bonneville Power Administration statement on added load at existing interconnections would release P-1 early.

Three events would break the set. Federal preemption of state data center rules would remove the county forum P-2 depends on. A hyperscaler acquiring regional anchor sites below 50 megawatts would collapse the customer tier. A Nuclear Regulatory Commission schedule slip past 2034 would void S-3’s window without changing the hydro route.

VII. Strategic Insights by Audience

Every audience faces the same first question in a different vocabulary: how much of this site’s power runs every hour. Policymakers meet it in allocation and accessory-use definitions. Executives meet it in site selection and hosting terms. Counsel meets it in import-rights and co-location clauses, and investors meet it in staging.

Strategic Posture by Audience

🏛️ Policymakers. Public utility districts and the Bonneville Power Administration hold the allocation lever, and counties hold the accessory-use determination. A definition keyed to firm import capacity and measured burden reaches the anchor tier that label-based rules strand. The small modular reactor site is the first nuclear unit sized to that definition.

The federal co-location question is the one the state cannot answer alone (P-1, S-3, S-4).

Sequence is what the simulation adds for policymakers. Behavioral Economics predicts that a committee responds to the salient project, which in Washington is the reactor, and Game Theory shows that a nuclear-only definition leaves the hydro anchor in negotiation at the site where the first agreement is likeliest. A committee that wants the first agreement writes the definition for the source that moves first, and the first bill text of the 2027 session shows which source it chose.

💼 Executives. Firm clean developers hold the anchor role and should sell it as regional inference plus a research tenancy. Variable clean developers hold the module role and should sell interruptible compute to the mid-scale tier. Reactor hosts should resolve the federal co-location question before any module ships (P-1, S-3).

Executives should read the anchor as a competitor to metro colocation, not to campuses. The sales motion then runs on firmness and burden instead of scale. Hyperscalers leave the tier alone because 20 megawatts does not move their capacity plans, and that absence is the developer’s market. The first host to file a burden exhibit writes the county’s template, and a template favors whoever wrote it.

⚖️ Counsel. The anchor turns on two clauses the current hosting market leaves silent: import rights at module load and allocation of approval delay. Nuclear hosts add the federal co-location question, which belongs in the service agreement before any operator agreement. Phase thresholds at 20 megawatts belong in the host agreement as named triggers (P-1, S-3).

For counsel, the binding clause changes with the source. At hydro hosts the agreement turns on import rights and restudy triggers, at reactor hosts on federal co-location cost allocation, and at variable hosts on curtailment terms and checkpoint duties. One template across the three sources imports the wrong risk into two of them.

📊 Investors. Underwrite the anchor on regional inference and interruptible compute. Stage capital to the executed allocation agreement at hydro sites and to the construction permit at reactor sites. Treat the research tenancy as contracted revenue only once signed (P-1, S-3).

Allocation risk and permitting risk resolve on different dates, and investors should hold them in separate gates. A county approval before an allocation agreement produces nothing to fund, and an allocation agreement before a county decision funds a site whose schedule the county still controls. Capital that treats the two as one gate pays for the slower of them twice.

Where Audience Interests Collide

Hosts and allocation authorities want different instruments. A host wants an allocation rule it can file against, and a district short of transmission wants case-by-case negotiation that preserves its discretion until the 2029 lines arrive. The 2027 definition of added load decides which instrument governs, and P-1 resolves in the host’s favor only where the district accepts a rule or the host accepts negotiation terms.

Developers and counties disagree on the test. A developer wants accessory-use treatment keyed to measured burden, and a county wants a reviewable standard it can apply to the next applicant. The first burden exhibit satisfies both only if it arrives before a pause, which is why P-2 turns on who files first.

Investors and reactor hosts run on different clocks. A reactor host wants an early anchor commitment to firm its offtake case, and an investor wants funding staged to the construction permit. The federal co-location docket decides whose schedule governs, and S-3 reads that docket as the gate.

Takeaway. Each audience’s best move follows from the firm fraction of the site in front of it, and the collisions resolve on dates the Simulation Predictions already carry.

VIII. Risk Mitigation by Simulation Prediction

Every entry below pairs a Simulation Prediction with the exposure it creates for one audience, the actions that audience can take alone and the exposure that survives. Severity and probability run on separate axes throughout.

Mitigating Actions by Time Gate

• Now to 30 days: classify every candidate site as anchor or module by firm fraction before any county or lender conversation.

• 30 to 90 days: obtain written confirmation of firm import capacity at anchor load from the public utility district or utility before signing a hosting agreement.

• 3 to 12 months: file burden exhibits on water, noise and generator hours with the first local filing, and prepare a draft accessory-use definition for the 2027 session.

• Beyond 12 months: keep anchor designs that satisfy both the accessory-use route and the large-load route, in case a county reclassifies or a phase crosses 20 megawatts.

Exposure, Actions and Residual for Each Simulation Prediction

The actions are analytic options for each audience and not legal, investment or fiduciary advice to any party.

Entry 1 · 💼 Host executives · P-1 (45–57%).

• Exposure: a host that signs an operator agreement before the public utility district confirms import capacity carries equipment deposits and quarters of lead time against an allocation that may not arrive. Severity: high. Probability: moderate.

• Mitigating actions: the development lead obtains written confirmation of import capacity at module load before signing, by the next district board cycle. The finance lead stages deposits against the executed service agreement. The commercial lead qualifies one geothermal or out-of-state hydro site as an alternative by mid-2027.

• Residual exposure: a district can revise allocation rules after the agreement, and a later phase can trigger restudy.

Entry 2 · 🏛️ Washington policymakers · P-1 (45–57%).

• Exposure: a district that allocates case by case sets the terms for every later anchor through negotiation instead of rule, and the legislature then inherits a precedent it did not write. Severity: moderate. Probability: moderate.

• Mitigating actions: legislative staff draft a definition of added load at existing generation interconnections before the 2027 bill cutoff. District staff publish an allocation test keyed to firm import capacity and measured burden before the first host files.

• Residual exposure: Bonneville Power Administration terms sit under federal authority and can override district tests.

Entry 3 · 💼 Developer executives · P-2 (42–54%).

• Exposure: a developer that seeks county approval for a quantum-equipped campus addition before any anchor has set the template loses the permitting schedule when the county has no accessory-use precedent to read. Severity: moderate. Probability: moderate.

• Mitigating actions: the development lead files the first quantum-equipped project at a firm-clean site under 20 megawatts rather than at a campus, through December 2028. The government affairs lead frames the equipment by function, recovery time and measured burden in the first filing.

• Residual exposure: a county may still classify the whole site by facility label during a pause.

Entry 4 · 🏛️ County policymakers · P-2 (42–54%).

• Exposure: a county that decides the first quantum-equipped project without an accessory-use test sets a precedent every later project will cite within one or two session calendars. Severity: high. Probability: moderate.

• Mitigating actions: planning staff draft burden thresholds for computing equipment at energy facilities in the next code amendment cycle. The board requests burden data at pre-application by resolution in the first quarter of 2027.

• Residual exposure: a statute passed under pro-development pressure can override county tests.

Entry 5 · ⚖️ Counsel at nuclear hosts · S-3 (40–52%).

• Exposure: a service agreement at a reactor site that is silent on federal co-location cost allocation leaves the host exposed to transmission charges a later FERC order assigns. Severity: high. Probability: moderate.

• Mitigating actions: regulatory counsel documents the client’s assessment of the pending co-location docket before execution. Transaction counsel adds a co-location reopener to every reactor-site agreement before signing.

• Residual exposure: FERC can adopt terms neither party proposed.

Entry 6 · 📊 Investors · S-3 (40–52%).

• Exposure: capital committed to a reactor-sited anchor ahead of the construction permit carries the Nuclear Regulatory Commission’s schedule as uncompensated delay, with operation in the mid-2030s. Severity: high. Probability: moderate.

• Mitigating actions: the investment committee stages funding to the construction permit and to a separate hydro or grid service agreement that firms the first phase. The deal team models the anchor on the interim source until the reactor operates.

• Residual exposure: the interim source may carry its own allocation limit.

Takeaway. Classifying the site by firm fraction protects every audience under both outcomes of P-1, because a module still earns if the allocation stalls and an anchor still earns if it arrives.

IX. Implications Run From Allocation to Tenancy

Public utility districts and utilities hold the binding economic determination at every anchor, because firm import capacity at anchor load is the input no county can grant. A county may act first in time, and its approval without import rights produces nothing. Counties decide the accessory-use question for anchors under 20 megawatts (P-2), and at nuclear hosts the federal co-location docket sits ahead of both. The sequence inverts the usual developer intuition that land use is the decisive approval.

Geothermal and hydro hosts capture the first research tenants, because they alone firm the anchor at modular scale among clean sources. Variable hosts capture the interruptible compute market and gain bargaining power by writing the first module terms. Reactor hosts capture the largest anchors and carry the longest schedule.

If P-1 misses, no Washington allocation authority executes an anchor agreement by 2028. The anchor model then moves to geothermal regions first, and Washington’s advantage in latency and research tenancy waits on a legislative definition. If P-2 misses, the first county decision lands on a campus, and the accessory-use template forms under campus burdens.

Takeaway. Allocation binds and county review sets the schedule. A miss at either Primary Simulation Prediction delays the anchor without changing the compute-yield table.

What to Watch: The 2027 Session, the Grant Lines and the First Agreement

The dominant fork is whether a Washington allocation authority executes an anchor agreement before geothermal regions build first (P-1).

• January 11, 2027. The Washington legislature convenes on the second Monday of January; bill text on added load, accessory use and tax classification moves S-4 and P-1.

• 2027. Grant County’s first transmission relief line energizes; headroom under the cap moves P-1 and S-2.

• Each FERC co-location order. Cost assignment at nuclear plants moves S-3 and the Richland reactor door.

• December 31, 2027. S-4 resolves against Washington statutes, rules and ordinances.

• December 31, 2028. P-1 and S-2 resolve against district minutes, utility filings and operator releases.

• 2029. Grant County’s second line energizes; P-2 resolves December 31 against the eighteen county dockets.

• December 31, 2030. S-1 and S-3 resolve against host, operator and reactor releases.

Verification sources by Simulation Prediction.

• Resolves P-1: public utility district board minutes and utility filings on added load at existing interconnections, reviewed monthly.

• Resolves P-2: county dockets listing quantum equipment, reviewed quarterly.

• Resolves S-1 and S-3: anchor announcements naming the host, the source and the operator, reviewed quarterly.

• Resolves S-2: inference contracts at Columbia River hydro sites, reviewed quarterly in operator releases.

• Resolves S-4: Washington statutes, agency rules and ordinances, reviewed during the 2027 session.

• Moves watch conditions: FERC co-location orders and Bonneville Power Administration statements, reviewed as they appear.

Conclusion: Build the Anchor on the Source, and Treat Allocation as the Binding Constraint

Clean energy source decides how much compute a modular site can host, and only geothermal, hydro and the first small modular reactors firm an anchor at a scale a county can approve as accessory use. Washington holds all three sources within the latency radius of two metros and a national laboratory. The public utility district allocation is the binding constraint there whichever authority acts first, and the legislature’s accessory-use definition decides whether that allocation runs by rule or by negotiation and whether the 2026 incentive change reaches anchors at all under the governing classification.

Stakeholders should act now on what they can measure: firm fraction at every candidate site, import capacity at anchor load and the terms of the first hosting agreement. Stakeholders can wait to learn whether the reactor schedule holds and whether a county reads the first anchor as accessory use.

The decisive signal is the first executed service agreement at an existing Washington hydro or nuclear interconnection. If it arrives at a first phase under 20 megawatts by 2028, P-1 holds and the anchor route governs. If it does not, P-1 misses and the model expects geothermal regions to build first while Washington waits on its definition.

Working With MindCast

Public analysis cannot tell a specific host whether its interconnection carries import at anchor load, whether its public utility district will allocate, or whether its county reads the project as accessory use. The answers depend on the site’s service agreement, the district’s contracted position and the county’s code. Site inputs stay confidential.

MindCast builds Cognitive Digital Twins of the host and its allocation authority. The county, the lender and the candidate operators join the same simulation, which ingests the organization’s own decision history and reruns the allocation contest against its actual options.

Commissioned scope adds complete probability distributions, P10/P50/P90 ranges, organization-specific calibration and complete scenario trees. The scope also adds the compute-yield model run on the site’s own hourly generation data, the fixed-versus-reconfigurable contest run on the operator’s own workload traces and module exchange costs, and red-team tests of district and county responses.

Three engagements map to this installment’s readers. An Anchor Diagnostic for a clean energy host: one site, one allocation authority and one configuration decision (P-1, P-2).

A Session Brief for a committee: the four 2027 instruments run against the district, county and host Cognitive Digital Twins (S-4). A Hosting Terms Review for counsel tests import rights and delay allocation together with phase thresholds and tax classification against the simulation’s counterparty responses (P-1, S-3). Contact mcai@mindcast-ai.com or visit www.mindcast-ai-simulation.com to arrange one.

Appendix A. MindCast Works

Companions

What Quantum Computing Contributes to Clean Compute Clusters, Tenant First, Training Data Second and Not the Processor Through 2032 (Installment Two). States what the anchor’s quantum island contributes by tier and date. Supplies the training-data loop that gives the synthesis workload its value.

How Quantum Improves AI, Five Channels Ranked by Date and Value, and the One That Pays for a Clean Compute Anchor (Installment Three). Ranks the channels through which quantum improves AI and finds that product demand does not create processor demand. Supplies the reason the anchor’s island circuit stays reserved rather than built.

Where the Clean Energy Host Advantage Holds and Where It Breaks in AI Data Centers (October 2026). Cover letter with seven Simulation Predictions on host sites. Supplies the five determinations, the finding that county review holds at energy sites and the hyperscaler hosting forecast this installment’s customer analysis relies on.

AI Data Centers and Clean Energy: Which Projects Reach Operation and Where Clean Technology Companies Can Win(October 2026). The magazine companion. Supplies the segment guide that this installment extends to anchor and module roles.

How Quantum Technology Changes the AI Data Center Energy, Compute and Environmental Calculus (October 2026). The quantum series overview. Supplies the quantum island and the finding that quantum shifts constraints rather than relieving them.

Funded Solutions That Can Help Win Quantum-Coupled AI Data Centers Approvable (October 2026). Tests 31 measures across three project types. Supplies the hub, addition and campus approval shares that bound the anchor’s quantum reservation.

The series umbrella and the market

AI Data Center Authorization + Clean Energy, the Mid-Scale Squeeze and the Specialized Compute Landscape(September 2026). The series umbrella. Supplies the Mid-Scale Squeeze and the Authorized Compute Portfolio that this installment instantiates with named sources.

AI Datacenter Edge Computing: Ship the Workload Not the Power (September 2025). The founding principle of moving compute toward energy and community trust. Supplies the siting logic the regional anchor puts into physical form.

The Grid-Anchored Clean Power Bargain: Powering AI Data Centers Through 2040 (September 2026). Separates interconnection, deliverable energy and accredited capacity. Supplies the firmness inversion that keeps the anchor on firm clean power.

The Clean Compute Match: Which Clean Power Pairs With Which AI Compute, and What Stays on Fossil (September 2026). Sorts compute and clean generation into matched pairs. Supplies the assortative matching rule the compute-yield table converts into GPU counts.

The approval layer

Burden-Based Authorization for AI Data Centers, Modular Clean Compute, and the Backlash Gap Between Grid Tariffs and State Law (September 2026). Defines the four burden measures. Supplies the measured-burden test the accessory-use anchor must pass.

AI Data Center Veto Points Show Where Projects Get Stopped and How Authorization Risk Moves (September 2026). Maps ten veto nodes and four state arrangements. Supplies the node sequence that places allocation ahead of county review.

The Lowest-Exposure AI Data Center Approval Route for Developers, Hyperscalers and Investors (September 2026). Ties capital release to durable approvals. Supplies the staging rule behind the reactor-site entries in the risk layer.

The Data Center Authorization Market: A 50-State Regulatory Atlas (August 2026). State-by-state evidence on approval terms. Supplies the state detail this installment leaves to the Atlas while carrying the three-condition rule.

The Model Data Center Authorization Code (July 2026). Compares any state’s provisions by function. Supplies the common language for the accessory-use definition the installment asks the legislature to draft.

The assembly

Modular AI Data Center Specialists and the Integrated Offerings That Turn Small Sites Into Authorized Operating Capacity (September 2026). Identifies power-asset hosts as the party that carries small sites to operation. Supplies the host role that the anchor and module distinction refines by source.

AI Data Center Taxonomy for Decision-Making and Clean Energy, Four Categories That Decide Whether a Site Operates(September 2026). Replaces facility labels with four decision categories. Supplies the “can change” and “can prove” categories through which the module’s curtailment and the anchor’s firm fraction enter the approval decision.

Appendix B. External Sources

• Energy Northwest and X-energy. Cascade Advanced Energy Facility near Richland in Washington State. Up to twelve Xe-100 modules with Amazon investment announced October 2024. Schedule as stated in X-energy’s Form 424B4, 2026.

• Grant County Public Utility District. Data center load cap, April 2025, as reported by GovTech. Transmission relief lines in 2027 and 2029.

• Washington State Legislature. Session calendar. The 2027 regular session convenes January 11, 2027.

• FERC. Order of October 9, 2026 in Docket ER26-3515 rejecting PJM Interconnection’s large-load curtailment proposal, with Commissioner Rosner’s concurrence. Retail curtailment as state jurisdiction.

• Business Wire. “Energy Vault Breaks Ground on Powered AI Infrastructure Campus in Snyder, Texas.” July 27, 2026. Module count, water use and phase sizes.

• Office of the Texas Governor. Directive of September 21, 2026 halting data center environmental permits, as summarized by Holland & Knight.

• Duke University Nicholas Institute. “Rethinking Load Growth.” February 2025. Estimate of 76 gigawatts of new load absorbable at 0.25 percent curtailment.

• Open Compute Project. “Integrating Quantum Processing Units into Data Center Infrastructure.” White paper, 2025. Facility classes for the anchor’s reserved island.

• Washington State Legislature. Senate Bill 6231, 2026. Ends the data center sales and use tax exemption for refurbished facilities and replacement servers from July 1, 2026, as summarized by Leyton.

Companion work: How Quantum Technology Changes the AI Data Center Energy, Compute and Environmental Calculus

I. Physical Constraints Decide Which Agreements Can Work, and Incentives Decide Which Survive

Physical constraints decide which agreements to build an artificial intelligence (AI) data center can work. Funding, enforcement and counterparty incentives decide which of the feasible agreements survive. Quantum computing adds one new physical constraint to that bargain: equipment that cannot lose power without a costly recovery.

Thesis. Under the modeled conditions, tailored measures reduce quantum’s added approval burden at every scale and close most of it where the land objection is negotiable. Large-load tariffs and contracted curtailment anchor the selected packages for new service, and ride-through answers quantum’s continuity objection. Hard land restrictions, long-duration backup noise and on-site fossil generation remain beyond what any tested package resolves.

Findings by solution family.

• Grid cost and reliability. Tariffs and curtailment anchor the selected packages for new service. Curtailment matters most: in removal tests at large campuses, dropping it cuts the share of cases that sustain an agreement by 20 points. Some small additions proceed inside existing service with few new terms.

• Quantum continuity. Battery ride-through sized to the equipment’s tolerance restores most of the approvability quantum removes. A protected reserve keeps that protection intact when the battery also earns grid revenue.

• Land and community. Setbacks, enforceable community benefit agreements and bonds resolve negotiable land objections. Payments change political objections but leave physical ones in place, and only relocation resolves a hard restriction.

• Water, cryogens and emissions. Funding new water supply beats over-designing cooling where supply is the objection. Cryogen recovery needs an initial inventory to recover, and on-site fossil generation trades a grid objection for an emissions objection.

• Shared infrastructure and measurement. Quantum hubs help compatible workloads but concentrate failure risk. Metering saves no power. Verified measurement supports enforcement when contracts define eligibility, remedies and payment conditions.

Bottom line on the controversy. In the model, the industry holds measures that reduce quantum’s added approval burden, though a gap remains at hubs. Developers rarely adopt the load-bearing local measures on their own, so those measures arrive as county conditions, negotiated agreements or rules. Stacking every measure fails, because the cost and the emissions objection from on-site generation break the bargain.

Quantum’s effect on the share of modeled cases that sustain an agreement

Shares are sensitivity results across modeled cases, not approval probabilities at real sites. Under identical mitigation, ride-through closes quantum’s gap to 1 to 3 points at additions and campuses. At hubs a 15-point gap remains, because cryogen supply and hub land use still bind.

Why now. The Edison Electric Institute’s October 9, 2026 list counts 25 states with approved large-load tariffs and 8 with only pending ones. On the same day, the Federal Energy Regulatory Commission (FERC) rejected PJM Interconnection’s curtail-first plan, finding that it appeared to dictate which retail customers get curtailed. In a concurrence, Commissioner David Rosner urged state regulators to establish clear rules for which customers get curtailed first. Local opposition has already blocked or delayed $64 billion of projects, according to Data Center Watch.

What distinguishes the MindCast approach. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through the MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation (MP CDT FS). Each Cognitive Digital Twin (CDT) models how a named actor decides as terms and evidence change.

Behavioral economics supplies the decision rule: developers favor the measure that relieves the most salient delay, even when it creates a worse objection later. Game theory supplies the payoff structure: an agreement survives only when the developer beats its outside option and the utility and community each come out ahead. Predictive behavior emerges from the combination.

What the companion does. The series overview names the burdens quantum shifts. The companion tests 31 measures across three project types and assesses which ones resolve those burdens, who pays and who adopts them voluntarily. The companion releases one MindCast Foresight Simulation Prediction and withholds three conditions the evidence does not yet support.

Simulation Prediction highlights

• R3-S-1 · 60–75% · by end-2027. At least five of the eight states with only pending large-load tariffs on the October 9, 2026 Edison Electric Institute list approve one.

Stakeholder readout

• 🏛️ Policymakers. Large-load tariffs will likely spread to most pending states by 2027 (R3-S-1). Pair each tariff with contracted curtailment and deadline-tiered recovery, and require enforceable local terms as county conditions.

• 💼 Executives. Classify each site’s land objection as hard or negotiable before spending on mitigation. Budget tariff terms into every pro forma, and protect quantum equipment on its own circuit.

• ⚖️ Counsel. Clawbacks and bonds strengthen remedies and financial security behind community benefit promises. Draft service agreements that anticipate a tariff in the eight pending states (R3-S-1).

• 📊 Investors. Underwrite the selected package for each scale, not the full catalog of measures. Treat on-site fossil generation as an approval risk rather than a delivery fix.

II. State Tariffs Answer the Grid’s Cost Question While Counties Hold the Veto

State tariffs now answer the grid’s cost question, while counties and water districts keep the vetoes that decide whether a site gets built. A package therefore has to satisfy two forums at once, and the simulation’s counterparties reflect that split.

A. Large-Load Tariffs Have Become the Norm and Local Authority the Binding Veto

Large-load tariffs have moved from exception to norm. The Edison Electric Institute counted 23 states with approved tariffs by May 2026 and 25 by October 9. Tariffs such as AEP Ohio’s data center schedule set minimum demand charges, collateral and exit fees so that residential customers do not carry stranded costs.

Flexibility has become the condition for fast connection. Duke University’s Nicholas Institute estimated that U.S. grids could absorb 76 gigawatts of new load if those loads accept curtailment 0.25 percent of the time. FERC’s October 9 order found that PJM’s plan appeared to intrude on state authority over retail curtailment. Commissioner Rosner’s concurrence separately urged state regulators to set curtailment priorities.

Local authority has become the binding veto. Texas Governor Greg Abbott ordered the state environmental agency to halt data center permits on September 21, 2026. Counties and water districts can deny projects outright, while commissions and utilities attach conditions.

Takeaway. A statewide tariff narrows a developer’s utility cost exposure without clearing its local veto.

B. Eight Terms Separate What Can Work From What Survives

Eight terms carry the assessment, and the split between hard and negotiable land does the most work.

• Large-load tariff: utility terms requiring a large customer to cover the costs it causes through minimum charges, collateral and exit fees.

• Deadline-tiered recovery: curtailment terms that schedule deferred work by task deadline, so curtailed load does not return as a new peak.

• Non-firm service: grid service the utility may interrupt until network upgrades are built.

• Ride-through: on-site storage that carries protected quantum equipment through a grid interruption.

• Hard land restriction: a zoning, environmental or parcel limit that no payment or agreement can remove.

• Negotiable land objection: a land-use concern that setbacks, enforceable benefits and bonds can reduce.

• Sustained agreement: a modeled case in which every holder of a denial right approves and the developer, utility and community each gain.

• Selected package: the set of measures the simulation’s search kept for a project type.

C. The Simulation Asks Which Funded Measures Keep Quantum Sites Approvable

The simulation answers one question. Which funded measures keep quantum-coupled AI data centers approvable, who pays for them, and which constraints remain after every tested measure?

The test covers three project types. Additions place a quantum system at an existing AI facility. Dedicated hubs house quantum systems for many tenants. New campuses run 200 to 500 megawatts.

The companion excludes the outcome of any specific pending permit. The model also leaves quantum’s computational value out, so each quantum-specific measure carries a cost that quantum value must cover.

Takeaway. The governing question separates what can work physically from what survives commercially and politically.

D. Salience Pulls Developers Toward the Wrong Fix

Developers anchor on the most visible delay. Grid connection queues dominate attention, so on-site generation looks like the fastest route to power. The simulation shows developers choosing on-site generation more often than any local measure at large campuses, even though it creates the emissions objection that breaks the agreement.

Counties discount promises they cannot enforce. A community benefit payment shifts political support, but authorities facing backlash give little weight to commitments without clawbacks or bonds. Authorities also decide under docket pressure, so the simulation models their choices with decision noise rather than perfect optimization.

Takeaway. Left to their own decision rule, developers favor the visible fix, so the measures that matter arrive through conditions or rules.

E. An Agreement Survives Only When All Three Parties Gain

An agreement requires three parties to gain at once. The developer must beat its outside option, the utility must recover its costs and the community must come out ahead. A measure that helps one party at another’s expense leaves the bargain short.

Mandates change choices without creating a business case. A parallel MindCast run found that packages meeting every technical condition can still exceed the project’s value, so the developer exits or redesigns instead of complying. Who captures each measure’s benefit therefore decides whether it gets adopted.

Takeaway. The three-party game decides which feasible packages survive, and the survivors depend on who funds each measure.

F. County and Water District Denial Rights Shape Every Modeled Package

The simulation models ten counterparties as Cognitive Digital Twins. The developer and host must beat an outside option, the hyperscaler tenant needs capacity on schedule and the quantum vendor needs systems that recover from outages. The utility serves load without stranded costs, and the state commission protects residential customers through tariff and curtailment authority.

Counties and water districts hold the only denial rights, over zoning and water service. Environmental reviewers condition emissions and noise, the community seeks local benefit and lenders protect repayment under terms modeled by stated assumption.

The run tested 31 measures in six families against twelve objections, from bills and reliability to cryogens, noise and hard land restrictions. Conditional effects reflect engineering limits. Retrofits need headroom, curtailment without recovery terms rebounds and new clean supply counts only when additional and on time.

The search selected a sustaining package for each project type and then removed every measure that added little. The search does not establish a global minimum, so a smaller or lower-cost package may exist. An overlap variant tests weaker combined effects by halving the second and later measures that act on one objection.

Section conclusion. Every tested package must pass a state forum on cost and a local forum on land, water and noise, and the developer must still beat its outside option.

III. Scale-Matched Packages Work Where the Full Catalog Fails

Scale decides the package. Inside each package, measures divide into those that carry the agreement, those that help only under conditions and those that backfire.

A. Scale Sets the Package, and Only Large Campuses Need Every Local Term

The search kept six measures at an ordinary addition and seventeen at a quantum campus.

• Addition, ordinary · 96% · $0.31M a year. Six measures: tariff, curtailment, local benefit, bond, setbacks and non-firm service.

• Addition, quantum · 95% · $0.97M a year. Seven measures: tariff, curtailment, local benefit, bond, ride-through, battery backup and helium measures.

• Hub, ordinary · 90% · $2.8M a year. Nine measures, adding brownfield siting, water supply funding and enforceable benefits to the ordinary addition’s six.

• Hub, quantum · 87% · $5.3M a year. Twelve measures: the ordinary hub package plus ride-through, helium measures and an industrial cryogenic plant.

• New campus, ordinary · 66% · $138M a year. Thirteen measures, listed below.

• New campus, quantum · 62% · $143M a year. Seventeen measures, listed below.

Each package was selected for its own project, so ordinary and quantum entries compare different packages. The quantum addition drops setbacks and non-firm service, which the ordinary addition keeps.

The large-campus package covers four groups:

• Grid and cost: tariff, curtailment, deadline-tiered recovery and non-firm service until upgrades.

• Siting and local terms: brownfield siting where available, setbacks, local benefit payments, enforceable community benefit agreements and bonds.

• Water and backup: water supply funding and battery backup.

• Measurement: metering and verified metering.

The quantum campus drops standalone metering and adds five measures. Ride-through and a protected battery reserve count as two. Helium measures, a platform without helium-3 where possible and hub access for compatible workloads make up the other three.

Under the overlap variant, large-campus shares fall to about 48% for both versions. Small additions inside existing service sustain an agreement in about a fifth of cases with no new measures at all.

Takeaway. Over-mitigating a small site costs approval: at an ordinary addition, the broad package sustains 54% of cases against 96% for the selected one.

B. Ride-Through Closes Quantum’s Gap Except Where Cryogen Supply Binds

Two comparisons answer different questions. Identical mitigation isolates quantum’s effect by applying the ordinary package to both sites. Tailored mitigation shows the residual gap after each site gets its own package.

• Addition. Under identical mitigation plus ride-through, the quantum site sustains 93% against 96%. Under tailored mitigation, the figures are 95% against 96%, and the whole quantum package costs about $0.7M a year more.

• Dedicated hub. Under identical mitigation plus ride-through, the quantum hub sustains 76% against 91%. Under tailored mitigation, the figures are 87% against 90%, at about $2.5M a year more.

• New campus. Under identical mitigation plus ride-through, the quantum campus sustains 63% against 64%. Under tailored mitigation, the figures are 62% against 66%, at about $5M a year more.

Without ride-through, quantum cuts the sustained share by 39 to 60 points at every scale. Ride-through costs about $0.1 million a year, annualized, at an addition and restores nearly all of the gap. At hubs, ride-through alone leaves 15 points open until helium measures and an industrial cryogenic plant join the package.

Takeaway. Hosts should budget ride-through before a quantum tenant arrives, and hub operators should secure cryogen supply before counting on ride-through.

C. Tariffs Decide Who Pays and Curtailment Decides When the Grid Can Serve

Grid measures divide into terms that decide who pays and terms that decide when the grid can serve.

• Load-bearing: the large-load tariff, paid by the developer, resolves utility cost recovery but not delivery timing. Contracted curtailment, paid by the hyperscaler, carries reliability everywhere. Deadline-tiered recovery keeps deferred work from returning as a peak, and non-firm service until upgrades resolves deliverability at a fraction of new supply’s cost.

• Conditional: new firm clean supply counts only when additional and on time. Reassignment to a qualified successor needs utility consent and real successor demand.

• Weak before 2031: a credit for released capacity rarely changes outcomes and invites speculative bookings unless eligibility rules and verified value gate it.

Curtailment carries the most weight of any single measure. In removal tests on the broad large-campus package, dropping it cuts the approvable share by 41 points. Without deadline-tiered recovery, curtailment keeps only 50 to 80% of its reliability relief, because deferred work returns as a new peak.

A parallel MindCast run quantified curtailment rebound directly. Deferring 80 megawatt-hours of work into two hours creates a 40-megawatt peak. A release credit set above booking cost also produced a positive return on speculative bookings. Eligibility rules, verified value and retained cost recovery guard against that return.

Takeaway. A tariff without curtailment leaves reliability open, and curtailment without recovery terms returns the load as a new peak.

D. Ride-Through Solves Quantum’s Distinctive Problem at Low Cost

Continuity is the one objection quantum adds on its own, and storage sized to the equipment answers most of it.

• Load-bearing: battery ride-through, paid by the host, carries refrigeration and control through short interruptions in every quantum package. A protected battery reserve keeps that energy available where the battery also earns grid revenue.

• Conditional: battery backup in place of diesel removes part of the noise from generator tests but not long-duration backup requirements.

• Supporting: a benchmark gate verifies complete-workflow claims before quantum scales, without touching physical burdens.

Ride-through lifts a quantum addition’s approvable share by 55 points on its own. Upfront cost runs about $0.72 million at an addition and $5.3 million at a hub under stated assumptions. Hosts adopt it voluntarily in 59% of modeled addition cases, the highest uptake of any measure.

Battery duration limits what ride-through and backup can cover. Where long-duration backup remains required, at 60 to 90% of hub and campus sites, batteries deliver half their noise relief. The parallel run found that an eight-hour battery leaves sixteen hours of a 24-hour event uncovered.

Takeaway. Hosts that sell battery capacity into grid markets need the protected reserve written into the hosting agreement.

E. Enforceability Moves Negotiable Land Objections, and Only Relocation Moves Hard Ones

Developers fund every land and community measure.

• Only fix for hard restrictions: brownfield or retired-plant siting, where such sites exist, also eases delivery and noise.

• Load-bearing at hubs and campuses: setbacks and noise walls resolve negotiable land use and noise where the parcel fits them. Enforceable benefit agreements with clawbacks address negotiable land use and local benefit, and bonds add financial security at low cost.

• Complementary: local benefit payments shift political opposition but leave physical burdens such as noise in place.

• Rarely needed: heat reuse helps only where a buyer and a heat network exist.

Hard restrictions appear at 15 to 35% of new campus sites under stated assumptions. Even the best package leaves them binding in 13% of ordinary and 17% of quantum campus cases. Setbacks, enforceable benefits and bonds resolve the negotiable share. Dropping setbacks in the large-campus removal tests cuts the approvable share by 15 to 16 points.

Developers almost never adopt local terms voluntarily: setbacks, enforceable benefits and bonds each raise developer value in under 5% of modeled cases. The parallel run found that heat reuse delivers nothing when the site has no heat buyer.

Takeaway. Counties gain leverage by requiring bonds and clawbacks, since developers rarely offer them unprompted.

F. Supply Funding Beats Over-Engineering, and On-Site Fossil Generation Backfires

Developers fund the water and generation measures, while vendors and hosts share cryogen duties.

• Load-bearing: water supply funding resolves supply objections where supply binds. A helium recovery and supply plan cuts cryogen losses at hubs when initial inventory exists.

• Conditional: reclaimed water changes the source of supply without cutting total use. A platform without helium-3 helps where the workload allows.

• Displaced: water-efficient cooling design cuts direct consumption but adds electricity demand, and supply funding resolves the same objection at lower cost.

• Counterproductive: on-site generation relieves grid delivery and creates an emissions objection.

Dropping water-efficient design in the large-campus removal tests raises the sustained share by 7 points, because supply funding resolves the same objection at lower cost. Water supply funding costs about $99 million upfront at a new campus under stated assumptions.

On-site generation breaks more agreements than it saves. With all 31 measures in place, an emissions objection appears in about 81% of cases. Developers still favor on-site generation at large campuses more often than any local measure.

Takeaway. A developer that answers grid delay with on-site fossil generation turns a solvable objection into one no tested package resolves.

G. Hubs Move Compatible Work, and Verified Measurement Supports Enforcement

Tenants pay for hub access, hosts for hub infrastructure and operators for metering.

• Load-bearing: hub access lets compatible quantum workloads leave the AI campus, though work needing tight coupling must stay. A shared industrial cryogenic plant secures supply at quantum hubs and adds a single point of failure.

• Helpful where available: a hub at a national lab or industrial park eases hub land use and cryogen supply.

• Enabling: metering resolves disputed consumption claims but saves no power. Verified metering with a public dashboard checks performance for claims and capacity credits, which contracts must still make enforceable.

Hub access lifts a quantum addition’s approvable share by 38 points. Hubs concentrate failure, however: the parallel run found one hub failure interrupting ten modeled customers. Adding a remote hub to a campus that needs tight coupling made the full package infeasible in four tested cases, and removing the hub restored feasibility.

Verified measurement supports enforcement when contracts define eligibility, remedies and payment conditions. Verified metering raises developer value in 20% of large-campus cases, more than any local term. The benchmark gate verifies complete-workflow benefits before a quantum system scales.

Takeaway. Hub siting should start from coupling needs, since a hub that breaks tight coupling makes the whole package infeasible.

H. Stacking Every Measure Collapses the Agreement

Stacking every measure collapses the agreement. With all 31 measures, the agreement holds in 3% or fewer of cases. The cost exceeds what the developer would accept, and on-site generation adds an emissions objection the community will not accept.

Broad packages also fail without on-site generation. A broad package that adds phased buildout and every local measure sustains 21% of large-campus cases, against 62 to 66% for the selected package. Measures a site does not need still carry their cost.

Takeaway. Matching the package to the site beats adopting the industry’s full catalog of measures.

Section conclusion. Four measures carry most of the weight across scales: tariffs, curtailment, ride-through at quantum sites and enforceable local terms where land is negotiable. Everything else helps only under site conditions, and on-site fossil generation hurts.

IV. Simulation Predictions: Large-Load Tariffs Spread to Most Pending States by 2027

The companion releases one Simulation Prediction, original to the companion and distinct from the overview’s set. States adopt tariffs at a measurable pace, so tariffs are the one solution the evidence lets the simulation forecast.

IV.A State Adoption Pace Makes Tariffs the Most Forecastable Solution

R3-S-1 | Large-Load Tariffs Spread to Pending States

Forecast statement: At least five of the eight states that had only pending large-load tariffs on the Edison Electric Institute list of October 9, 2026 approve a large-load tariff by December 31, 2027. Band: 60–75%.

• Pending states: Arizona, Delaware, Maryland, Montana, New Mexico, North Carolina, Oklahoma and South Dakota.

• Class and window: event, through December 31, 2027.

• Base rate: 23 states approved by May 2026, and 25 approved with 8 pending by October 9, 2026.

• Explicit falsifier: four or fewer of the eight states approve a large-load tariff by the deadline.

• Verification source: state commission orders and the Edison Electric Institute list. No order found counts as no approval.

R3-S-1 rests on its own event model. The approval shares in Section III are sensitivity results and do not inform its band.

Takeaway. Developers in the eight states should plan on a tariff as the base case rather than the downside.

IV.B Enforceable Local Terms Still Need County Evidence Before a Band

The run withheld bands for three conditions because the evidence did not support release. Each stays under watch and enters a later set of Simulation Predictions when evidence arrives.

• First quantum approval decision rests on ordinary burdens. The governing claim rates exploratory, since no county has yet decided a quantum-equipped project.

• First tariff that credits released reserved capacity. No precedent exists.

• First quantum hosting specification requiring metering by subsystem. None is published.

County decisions on large campuses where developers offered setbacks, enforceable benefits and bonds would support a large-campus forecast. Each decision should show the outcome, its grounds and whether the site carried a hard restriction.

IV.C State Authority Over Tariffs Anchors the Forecast

A federal rule setting large-load terms would break R3-S-1’s mechanism. FERC’s October 9 order treated retail curtailment terms as a matter of state jurisdiction, so the forecast assumes state commissions keep that authority through 2027.

The assessment rests on stated assumptions. Counterparty positions at the county, water district, utility and lender are modeled. So is the share of sites carrying hard restrictions. Costs are anchored to public data and need site quotations before they support a decision.

Section conclusion. The tariff forecast anchors the near term, while the large-campus finding waits on county decisions that would support a band.

V. Every Stakeholder Should Classify the Site, Fund the Anchors and Enforce Local Terms

Counties and commissions impose the terms, developers and hosts fund them, counsel drafts them and investors decide which packages to underwrite. Exposure therefore lands on each party at a different point in the sequence.

Counties Gain Most by Requiring What Developers Will Not Volunteer

🏛️ Policymakers. State commissions hold the tariff and curtailment levers, while counties hold the denial right. Tariffs answer residential bill exposure, and curtailment with deadline-tiered recovery answers reliability. Counties gain the most by requiring setbacks, enforceable benefits and bonds as conditions, since developers rarely offer them voluntarily.

💼 Executives. Developers, hosts and hyperscalers share one sequence. Classify the land objection first. Then budget tariffs, curtailment and non-firm service, and add the local terms the site needs. Hosts protect quantum equipment with ride-through and a protected reserve before the first tenant installs.

⚖️ Counsel. Enforceability separates a community benefit from a payment. Clawbacks and bonds strengthen remedies and financial security, while verified metering lets counties check performance. Service agreements signed in the eight pending states should anticipate tariff terms (R3-S-1).

📊 Investors. The selected package by scale is the right underwriting case. Selected quantum packages cost about $0.7 million a year more at an addition and $2.5 million more at a hub, which quantum value must cover. Hard land restrictions and on-site generation mark the cases where no package sustains the agreement.

Measured Gaps Become Documented Knowledge Once a Party Finds Them

MindCast’s Foresight-to-Duty Conversion in AI Governance (2026) shows how an institution’s own measurement becomes evidence once it keeps the gap it found. The solutions assessment produces such measurements early: land classification, ride-through sizing, rebound schedules and cryogen inventory.

Judge Learned Hand’s formulation weighs the burden of precaution against the probability and magnitude of harm together. Several tested measures carry small modeled burdens, such as ride-through at about $0.1 million a year at an addition. A low modeled mitigation cost alone does not complete that analysis, because probability and magnitude depend on the site.

• Hard land restriction found after site commitment. The developer’s site team holds the measurement and the developer funds the fix. County planning staff verify it.

• Quantum equipment without sized ride-through. Host engineering holds the measurement and the host funds the fix. Vendor recovery tests and the tenant verify it.

• Curtailment without a catch-up schedule. Hyperscaler operations hold the measurement and the hyperscaler funds the fix. The utility verifies it.

• Community promise without clawbacks or bond. The developer’s community team holds the measurement and the developer funds the fix. The county and community verify it.

• Cryogen plan without confirmed initial inventory. The vendor and host hold the measurement and the vendor funds the fix. The tenant and host verify it.

• On-site generation proposed as a delivery fix. The developer’s energy team holds the measurement and the developer funds the fix. Environmental reviewers verify it.

Takeaway. Documented awareness can strengthen the case for evaluating and addressing a foreseeable risk.

Site Classification Comes First and Quantum Continuity Terms Follow

The sequence starts with site classification, because every later spend depends on whether the land objection is hard.

• Now to 30 days: classify every pipeline site as hard or negotiable before a county cites a non-negotiable standard.

• 30 to 90 days: add clawbacks and bonds to pending benefit agreements before a county requests enforceable terms.

• 3 to 12 months: model tariff terms in the eight pending states before a pending tariff reaches a commission vote.

• Beyond 12 months: write ride-through and reserve duties into hosting agreements before a quantum tenant signs a letter of intent.

Tariff Spread Exposes Budgets, Dockets, Agreements and Underwriting

Entry 1 · 💼 Developers · R3-S-1 (60–75%).

• Exposure: a developer that budgets a large campus in one of the eight states at standard service terms omits a modeled $58 million upfront and $27 million annual tariff-and-upgrade obligation if its state adopts comparable terms. Severity: high. Probability: R3-S-1 forecasts the eight-state count only and supplies no probability for any single state or developer.

• Mitigating actions: the development finance lead models tariff terms in every pro forma for the eight states before the next interconnection filing. The commercial lead negotiates minimum-demand and exit terms before the state’s tariff order. The site lead ranks pipeline sites by tariff exposure through December 31, 2027.

• Residual exposure: final tariff terms vary by state and may exceed the modeled obligation.

Entry 2 · 🏛️ Policymakers · R3-S-1 (60–75%).

• Exposure: a pending state that has not approved a tariff by end-2027 leaves residential customers exposed to stranded upgrade costs for at least one more session calendar. Severity: high. Probability: R3-S-1 forecasts the eight-state count only and supplies no probability for any single state.

• Mitigating actions: commission staff publish draft terms with minimum charges, collateral and exit fees before the 2027 sessions. Staff pair the tariff with curtailment priorities, as Commissioner Rosner’s concurrence urged state regulators to establish.

• Residual exposure: large customers can contest terms and delay a final order.

Entry 3 · ⚖️ Counsel · R3-S-1 (60–75%).

• Exposure: a service agreement executed in a pending state without tariff-change provisions widens the dispute over which terms apply once a tariff takes effect. Severity: moderate. Probability: R3-S-1 forecasts the eight-state count only and supplies no probability for any single state or agreement.

• Mitigating actions: transaction counsel adds tariff-change provisions to agreements executed before each state’s order. Regulatory counsel documents the client’s assessment of pending docket terms before execution.

• Residual exposure: a commission may apply new terms to agreements signed during the pending docket.

Entry 4 · 📊 Investors · R3-S-1 (60–75%).

• Exposure: an underwriting case using standard service terms in the eight states overstates annual margin by the tariff cost, about $0.94 million at a hub and $27 million at a large campus under stated assumptions. Severity: moderate to high. Probability: R3-S-1 forecasts the eight-state count only and supplies no probability for any single state or asset.

• Mitigating actions: the credit committee applies tariff terms to the base case in all eight states through 2027. The diligence lead requires the utility’s draft terms before closing.

• Residual exposure: a state that adopts unusually strict terms can still compress returns below the modeled case.

Nothing in Section V constitutes legal, investment or fiduciary advice.

Section conclusion. The tariff forecast gives every party the same near-term action: build tariff terms into budgets, dockets, agreements and underwriting before the eight pending orders arrive.

VI. Local Enforceability and State Tariffs Divide the Work

Counties and water districts decide large campuses, while developers and hyperscalers fund most of what those authorities require. The split shapes institutions, distribution and the consequences of a missed tariff forecast.

A. County Conditions Will Shape More Outcomes Than Statewide Rules

Counties and water districts hold the vetoes that decide large campuses. State tariffs resolve utility cost exposure but leave land, noise and water with local authorities. Local conditions requiring enforceable terms will therefore shape more outcomes than any statewide rule.

B. Developers Fund the Local Terms and Hosts Fund Continuity

Developers fund most local measures, while hyperscalers fund curtailment and non-firm service. Hosts fund ride-through and gain the early quantum tenants. Residents gain bill protection from tariffs and local terms from enforceable agreements.

C. A Missed Forecast Could Leave Utility Cost Terms to Case-by-Case Negotiation

If R3-S-1 fails, four or fewer of the eight pending states approved a tariff by the deadline. Utility cost terms could then continue through case-by-case negotiation. The assessment’s other findings would still hold, since they rest on sensitivity results rather than on the forecast.

Section conclusion. State tariffs and local enforceability divide the work, and neither substitutes for the other.

VII. Each Installment Inherits a Tested Measure and an Open Question

Each of the five installments inherits a tested measure from the companion and an open question it must answer with its own simulation.

• Power grid: curtailment, recovery terms and non-firm service arrive as load-bearing measures. The grid installment will test whether flexibility terms change reserved capacity and peak.

• Cross-border compute: hub access carries additions, while hubs concentrate failure. The cross-border installment will test where compatible quantum workloads move under export controls.

• Clean technology market: ride-through, protected reserves and cryogenic plants become purchased services. The clean technology installment will test what clean developers and storage suppliers can sell.

• Environmental balance: on-site generation rates counterproductive, and supply funding beats over-design. The environmental installment will test total burden by architecture, with rebound.

• Data center approval: the split between hard and negotiable land carries forward, along with enforceable county conditions. The approval installment will test county decisions on quantum-equipped projects and the credibility of claims.

Section conclusion. The approval finale inherits the most, because the companion’s central claim on enforceable local terms still awaits county evidence.

Tariff Orders in Eight States Resolve the Forecast First

Four signals resolve or move the companion’s forecast and watch conditions.

• Resolves R3-S-1: large-load tariff orders in the eight pending states, reviewed monthly in commission dockets and the Edison Electric Institute list.

• Breaks R3-S-1: a federal large-load rule, reviewed at each FERC order.

• Supports a large-campus forecast: county decisions on campuses offering enforceable terms, reviewed quarterly.

• Releases watch conditions: tariffs crediting released reserved capacity and quantum hosting specifications with subsystem metering, reviewed quarterly.

Section conclusion. Monthly tariff orders resolve R3-S-1, while quarterly county decisions decide whether a large-campus forecast follows.

Conclusion: Counties and Commissions Decide Whether the Measures That Matter Arrive

Physical constraints decide which agreements can work, and funding, enforcement and counterparty incentives decide which of them survive. Large-load tariffs and contracted curtailment anchor the selected packages for new service. Ride-through answers quantum’s continuity objection at low cost, while cryogen supply keeps a gap open at hubs.

In the model, large campuses become approvable more often where the land objection is negotiable and the county can enforce the terms. Hard land restrictions, long-duration backup noise and on-site generation remain the cases no tested package resolves. Developers rarely volunteer the measures that matter most, so counties and commissions will decide whether those measures arrive.

The decisive near-term signal is tariff adoption in the eight pending states. If four or fewer approve a large-load tariff by December 31, 2027, R3-S-1 fails. Utility cost terms could then continue through case-by-case negotiation.

A Site’s Package Depends on Inputs Public Analysis Cannot See

Public analysis cannot tell a specific developer which package its site needs. The answer depends on whether the land objection is hard or negotiable, how much existing service covers and what the county will enforce. Site inputs stay confidential.

MindCast builds Cognitive Digital Twins of the developer and its counterparties: utility, county, water district and prospective quantum tenants. The engagement reruns the package search against the site’s actual options, costs and counterparties.

Commissioned scope adds complete probability distributions, P10/P50/P90 ranges and organization-specific calibration. The scope also adds package alternatives by cost, removal tests on every measure and red-team tests of county and utility responses.

The bounded next step is a scoped site diagnostic: one site, one county and one package decision. Contact mcai@mindcast-ai.com or visit www.mindcast-ai-simulation.com to arrange it.

Appendix A. Selected MindCast Works

The series

• How Quantum Technology Changes the AI Data Center Energy, Compute and Environmental Calculus (October 2026). The series overview. Names the burdens quantum shifts across power, cross-border compute, clean technology, environment and approval, which the companion tests solutions against.

Frameworks and methods

• The Dynamic Predictive Game Theory Collection (July 2026). Supplies the changing-game frame used to model the three-party bargain among developer, utility and community.

• The Authorization Market: Standardized Bargaining, Rationed Power, and the Competition to Build America’s AI Infrastructure (August 2026). Supplies the finding that connected power is not firm power, which frames non-firm service until upgrades.

• Burden-Based Authorization for AI Data Centers, Modular Clean Compute, and the Backlash Gap Between Grid Tariffs and State Law (September 2026). Supplies the measured-burden test behind the split between tariffs and local conditions.

• Foresight-to-Duty Conversion in AI Governance, Series Introduction (October 2026). Supplies the conversion of measured gaps into evidence of awareness, the frame for the Foresight-to-Duty list.

Related AI data center analysis

• AI Data Center Veto Points Show Where Projects Get Stopped and How Authorization Risk Moves (September 2026). Maps the county and water district vetoes that the selected packages must clear.

• The Lowest-Exposure AI Data Center Approval Route for Developers, Hyperscalers and Investors (September 2026). Supplies the route-selection logic that the scale-matched packages extend.

• The Grid-Anchored Clean Power Bargain: Powering AI Data Centers Through 2040 (September 2026). Supplies the firmness inversion that keeps quantum islands apart from curtailable campuses.

• The Clean Compute Match: Which Clean Power Pairs With Which AI Compute, And What Stays On Fossil(September 2026). Pairs firm clean power with anchored compute, the supply route the companion rates conditional.

Appendix B. External Sources

• Edison Electric Institute, “List of Large Customer Projects and Tariffs“ (updated October 9, 2026).

• FERC, Order on PJM Interconnection Large Load Proposal, Docket ER26-3515-000 (October 9, 2026).

• Utility Dive, “FERC Rejects Core of PJM Large Load Interconnection Plan“ (October 2026), reporting Commissioner Rosner’s concurrence.

• Data Center Watch, “$64 Billion of Data Center Projects Have Been Blocked or Delayed amid Local Opposition“ (2025).

• Bracewell, “Governor Abbott Orders TCEQ to Pause All Permit Issuance for Data Center Projects“ (September 23, 2026).

• AEP Ohio, Data Center Tariff (Schedule DCT).

• Columbia Law School Sabin Center, “Data Center Regulation: What Local Governments Should Know about Large-Load Tariffs and Clean Transition Tariffs“ (June 2, 2026).

• Duke University Nicholas Institute, “Rethinking Load Growth“ (February 2025).

• NREL Annual Technology Baseline, Commercial Battery Storage.

Companion work: Funded Solutions That Can Help Win Quantum-Coupled AI Data Centers Approvable

I. Quantum Shifts AI Data Center Constraints More Than It Relieves Them Through 2030

Quantum computing is arriving in AI data centers as small, continuously powered systems linked to classical computers. Through 2030, those systems mostly shift infrastructure constraints instead of relieving them. They add support loads, require protected power and move part of the burden into cryogenics and export-controlled locations.

Thesis. Quantum lowers site consumption when displaced classical energy exceeds supporting overhead and additional output stays below the resulting savings threshold. Reserved capacity can stay unchanged even when consumption falls, so the series tracks it separately. The simulation finds the first condition unproven at scale before 2031 and the second fragile wherever it holds.

Findings by channel.

• Power grid. Consumption relief is narrow and fragile to added output, while protected power for a quantum island becomes a siting requirement. A quantum island means the refrigeration, control electronics and host computer protected on their own circuit.

• Cross-border compute. Globally distributed access can connect locally concentrated computing islands. Real-time control stays within one building, while job-level access crosses continents under export controls.

• Clean technology market. The islands need continuity, cooling and recovery services more than energy volume.

• Environmental balance. Burden differs by architecture, with helium-3 exposure concentrated in dilution-refrigerated systems.

• Data center approval. Quantum can change approval prospects through measured changes in project burden and through the credibility of the developer’s claims.

• What helps now. Shared workloads across remote sites and specialized compute matched to clean power help AI data centers far more than quantum will before 2031. Section V.D sets out how much each helps now and later.

Bottom line on the controversy. Through 2030, quantum does not reduce the controversy surrounding AI data centers. Adding quantum without removing classical work raises power demand and environmental burden, though by small amounts at current scale. The larger effect runs through credibility: developers who promise quantum offsets will deepen the controversy, while developers who measure burden and protect a small subsystem will leave it unchanged.

Quantum reduces the controversy only if it avoids more burden than its complete infrastructure creates, after accounting for additional output.

Quantum’s impact on the AI data center controversy through 2030

Why now. On October 7, 2026, the Defense Advanced Research Projects Agency (DARPA) moved four companies into the final stage of its Quantum Benchmarking Initiative. The four are Atom Computing, Diraq, IBM and IonQ. DARPA will test whether any approach reaches utility scale, where computational value exceeds cost, by 2033. Two days later, the Federal Energy Regulatory Commission (FERC) rejected PJM Interconnection’s curtail-first plan for new large loads, finding that it appeared to dictate which retail customers get curtailed.

What distinguishes the MindCast approach. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through the MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation (MP CDT FS). Each Cognitive Digital Twin (CDT) models how a named actor decides as rules and evidence change.

Behavioral economics supplies the decision rule: authorities discount salient claims they cannot measure. Game theory supplies the payoff structure: where backup for a small subsystem is affordable, broad protection adds little for the operator. Predictive behavior emerges from the combination.

What the overview does. The overview releases three MindCast Foresight Simulation Predictions and withholds five conditions the evidence does not yet support. Section VI then maps five installments, from the power grid to data center approval.

Simulation Prediction highlights

Stakeholder readout

• 🏛️ Policymakers. When the first qualifying instrument addresses quantum continuity, the simulation expects it to require backup rather than grant an exemption (P-1). Before the 2027 sessions, define protected load by function and recovery time rather than technology label.

• 💼 Executives. Developers, hosts and hyperscalers gain an approvable design in the engineered quantum island (P-1, S-1). Isolate refrigeration and control on a protected circuit with sized ride-through before the first hosted installation.

• ⚖️ Counsel. Once a client measures a quantum gap, such as unprotected refrigeration or an efficiency claim without support-system data, the file becomes relevant evidence of awareness (P-1, P-2). Assess each gap case by case and document the decision before filing.

• 📊 Investors. If a site discloses measured relief, the simulation expects added output to absorb it (P-2). Give quantum efficiency no capacity credit until a measured disclosure exists.

II. Quantum Enters Data Centers as a Coupled Subsystem While Authorities Demand Measurement

Quantum enters AI data centers as a small coupled subsystem at the moment authorities demand present measurement from every large load. The collision matters more than quantum’s size, because forums under backlash pressure discount anything they cannot verify. Behavioral economics explains why actors overstate quantum’s efficiency, and game theory explains how state forums respond.

A. Demonstrations Still Outrun Deployments as Authorization Tightens

Commercial integration has begun at small scale. Equinix and Diraq announced Australia’s first quantum computer inside a data centre on September 1, 2026. The announced eight-qubit system would draw less than 20 kilowatts with self-contained cooling and controls.

Vendors have documented how to couple quantum processors to graphics processing units (GPUs). NVIDIA’s NVQLink integration architecture places a real-time classical host inside the quantum system for decoding and calibration. IonQ announced on September 23, 2026 that its Superion 256 will connect to an NVIDIA GB200 NVL72 system through NVQLink, with installation scheduled for 2027.

Demonstrations still outrun deployments. Quandela’s October 13 NVQLink demonstrator used a simulated quantum processor at an exhibition. Researchers have only begun to define energy efficiency across five quantum platforms, in Carrasco-Codina and colleagues’ “Energy Efficiency of Quantum Computers” (2026).

Authorization has tightened around large loads. Texas Governor Greg Abbott ordered the state environmental agency to halt data center permits on September 21, 2026. The halt holds pending an audit of the grid operator’s queue, expected in mid-December.

Takeaway. Quantum is moving into data centers as a coupled subsystem, while approval authorities demand present measurement from every large load.

B. Relief Counts Support Systems, and Reserved Capacity Moves Separately

Seven terms carry the energy accounting, and two of them separate what headlines merge: relief and reserved capacity.

• Quantum island: refrigeration, control electronics and host computer protected on their own circuit.

• Ride-through: on-site storage that carries protected equipment through a grid interruption.

• Curtailment: an order from a grid operator or utility that a load reduce or stop drawing power.

• Coupling: the real-time link between a quantum processor and its classical host.

• Relief: lower site energy consumption for equivalent useful work, counting support systems.

• Reserved capacity: grid capacity a site contracts for, tracked separately from consumption.

• Rebound: added output that absorbs energy savings, so total consumption holds.

Takeaway. Relief counts support systems, and reserved capacity moves separately from consumption, so a site can save energy without freeing grid capacity.

C. The Simulation Separates Computational Usefulness From Physical Relief

The simulation answers one question. Which quantum-hybrid deployments convert quantum capability into lower resource requirements, and how do operators, hosts and authorities respond?

Released Simulation Predictions run from October 2026 through 2031. Scenarios extend to 2040 without dated bands. The forum is U.S. federal and state authority, with non-U.S. hosting as a comparison case.

The overview excludes post-quantum cryptography, quantum sensing and the selection of individual securities.

Takeaway. The governing question separates computational usefulness from physical relief and both from institutional treatment.

D. Authorities Discount Quantum Claims They Cannot Measure

Narrative analysis treats quantum as an efficiency story. Behavioral economics shows why actors use that story selectively. Operators anchor capacity plans to firm megawatts, so efficiency claims serve as arguments rather than as reasons to retire capacity.

Salience matters in approval forums. A future quantum offset sounds compelling in testimony, yet authorities under backlash pressure discount claims they cannot measure. Authorities also decide under docket pressure with bounded attention, so the simulation models their choices with decision noise rather than perfect optimization.

Takeaway. Declared efficiency goals and revealed capacity behavior diverge, and the divergence drives the forecasts.

E. State Forums and Operator Learning Shape Quantum’s Continuity Terms

FERC’s October 9 order found that PJM’s plan appeared to intrude on state authority over retail curtailment. Curtailment treatment for quantum equipment therefore runs through state forums with their own pressures. A static estimate of quantum’s energy effect misses that forum.

Operators learn from authorities’ responses. An operator that requests broad protection and receives a backup duty updates its expectations and shifts to a subsystem request. Vendors hold the energy data, so information asymmetry shapes every early claim.

Takeaway. Game theory supplies the payoff structure and the selection among equilibria, and predictive behavior emerges from combining it with behavioral decision rules.

F. Eleven Named Actors Test Whether Quantum Adds Load, Substitutes or Waits

The simulation models eleven actors as Cognitive Digital Twins. Hyperscale clouds (Amazon Web Services, Microsoft and Google) seek firm capacity and sell quantum access under queue, curtailment and export limits. Equinix hosts quantum systems as a premium product. Diraq, IonQ and IBM Quantum pursue placement, coupled results and fault tolerance under refrigeration and schedule limits.

NVIDIA works to make GPUs part of every quantum system within a microsecond latency budget. On the public side, FERC defines wholesale large-load service and PJM admits large loads after FERC’s rejection. Texas authorities admit loads with curtailment and backup duties during a permit halt. The Department of Energy and DARPA verify utility-scale value by 2033, while project lenders protect repayment under terms modeled by stated assumption.

The run tested whether quantum substitutes for classical work, adds load without displacing it or stays too small to matter through 2031. Four executed models fed the test: complete-system energy accounting, continuity sizing, coupling latency and a repeated contest over continuity terms.

Takeaway. The results favor delayed usefulness, added load and subsystem continuity over substitution, while the too-small-to-matter route stays live.

Section conclusion. Quantum’s treatment will turn on measured burden and continuity terms set in state forums, not on its energy draw.

III. Simulation Predictions: Backup Duties Precede Exemptions and Added Output Absorbs Energy Savings

The overview releases three Simulation Predictions, original to the overview, and each tests a different link in the argument. P-1 tests how authorities treat quantum continuity, P-2 tests whether energy savings reach the grid and S-1 tests whether coupling arrives on schedule.

Small Protected Loads Keep Backup Modest While Added Output Erodes Relief

Energy accounting. Site consumption falls when displaced classical energy exceeds the energy of supporting systems: refrigeration, control electronics and the classical host. Added output then consumes part of the savings. Each deployment therefore has a growth threshold beyond which consumption rises, and in typical modeled cases the threshold sits at a few percent of added output.

Reserved capacity. A site can consume less energy while contracting the same grid capacity. The series treats consumption and reserved capacity as separate observations.

Continuity requirement. Protected energy scales with the protected load and the outage duration. Quantum systems are small, so the energy quantity stays modest, while cost and integration still require case-specific assessment.

Coupling distance. Real-time decoding requires a round trip of a few microseconds, which keeps a quantum processor and its classical host within a few hundred meters. Job-level access tolerates delays of tens to hundreds of milliseconds and can cross continents.

Continuity terms. The simulation modeled repeated negotiations between operators and authorities over continuity terms. Where backup for a small subsystem is affordable, broad protection adds little for the operator, and authorities facing measurement demands favor backup duties. Private hosts can require ride-through but cannot grant curtailment exemptions.

Takeaway. Small protected loads keep backup requirements modest in energy terms, while small eligible shares make consumption relief fragile to added output.

Delayed Usefulness and the Equipment Boundary Lead the Scenario Routes

Among physical outcomes, delayed usefulness is the base route: useful advantage at scale arrives after 2030 unless a complete-workflow advantage gets disclosed. Addition without displacement runs second, and measured substitution on a narrow workload remains the tail.

Among institutional outcomes, the equipment boundary is the base route as ride-through and backup duties spread to hosted installations. Generic treatment below the attention of dockets runs second, and a technology-label exemption adopted by a pro-development forum remains the tail.

III.A Continuity Attaches to the Subsystem and Relief Feeds Added Output

P-1 | Backup Duty Before Exemption

Forecast statement: P(the first instrument addressing quantum-equipment power continuity uses storage-backed ride-through or a backup obligation rather than a curtailment exemption or critical-load status given that such an instrument appears by year-end 2028). Band: 90–100%. An instrument means an agreement, tariff, order or published specification.

• Class and window: conditional event, through December 31, 2028; void if no qualifying instrument appears.

• Trigger conditions: hosted installations outside vendor-owned sites, with continuity requests reaching hosts or state forums.

• Explicit falsifier: the first instrument grants an exemption or critical-load status without a ride-through or backup duty.

• Verification source: the text of the first instrument, whether a tariff filing, service agreement, colocation specification or order.

P-2 | Added Output Absorbs Energy Savings

Forecast statement: P(the disclosing site’s useful output or utilization rises and its total metered consumption does not fall within twelve months given a measured relief disclosure by year-end 2030). Band: 50–60%.

• Class and window: conditional event, twelve months after the disclosure and through December 31, 2031; void if no qualifying disclosure appears by 2030.

• Trigger conditions: a named site discloses lower energy consumption for equivalent useful work, with support systems inside the boundary.

• Explicit falsifier: neither useful output nor utilization rises, or total metered consumption falls.

• Verification source: operator disclosures, service agreements and interconnection filings for the disclosing site.

Missing data counts against P-2. If the condition is met but output or consumption data stay undisclosed, P-2 resolves as a miss and an evidence gap.

III.B Coupling Milestones Test Integration, Not Energy Savings

S-1 | Superion 256 Coupling Confirmed

Forecast statement: IonQ or NVIDIA confirms completed installation of Superion 256 at NVIDIA’s Accelerated Quantum Research Center and its coupling to an NVIDIA GB200 NVL72 system through NVQLink by December 31, 2027. Band: 50–60%.

• Linked route: addition.

• Trigger: the announced 2027 installation.

• Explicit falsifier: no confirmation of completed installation and coupling by the deadline; a further installation announcement does not count.

S-1 tests a coupling milestone only. The outcome carries no claim about energy savings or aggregate GPU demand.

III.C Offset Claims and Relief Disclosures Remain Too Thin to Forecast

The run withheld dated bands for five conditions because the evidence did not support release. Each stays under watch and enters a later set of Simulation Predictions when a gating event or new evidence arrives.

• First measured relief disclosure by 2030. Timing evidence remains too thin for a band, and P-2 activates only if the disclosure occurs.

• First continuity instrument by 2028. Hosting pipelines exist, but the link from installation to a qualifying instrument lacks support. P-1 activates only if the instrument appears.

• Quantum offset citations in state proceedings. No qualifying filing has appeared, and actor intent remains unobserved.

• Quantum-inspired energy disclosures first. Vendor claims exist, but no named operator has disclosed measured kilowatt-hours.

• Quantum cited as grounds for curtailment exemption. No quantum installation yet sits inside a curtailable U.S. campus.

Takeaway. Offset claims and relief disclosures will reach state forums before the evidence to forecast their timing does.

III.D A Critical-Load Statute Would Break the Backup-Duty Forecast

One rule change breaks P-1. A state statute designating a whole facility class as critical load would grant protection without a quantum-specific contest. Such a statute counts against P-1 only if it names quantum equipment, and it shifts weight toward the broad-protection route regardless.

Lender behavior rests on a stated assumption. No lender has disclosed how it treats quantum efficiency, so the simulation assumes lenders credit capacity rather than unmeasured claims. If lenders credit announcements, P-2 weakens.

Section conclusion. P-1 at 90–100% tells developers and commissions which continuity design will hold, while P-2 at 50–60% tells investors not to count quantum savings as freed capacity. S-1 at 50–60% confirms integration progress only.

IV. Every Stakeholder Should Separate the Quantum Island From the Campus

Every stakeholder faces the same design choice in a different vocabulary: keep the quantum island separate from the campus or bundle the two. Policymakers make the choice in protected-load definitions, executives in electrical design, counsel in filings and investors in capacity credit.

Quantum Gives State Commissions a Clean Test Case for Protected Load

🏛️ Policymakers. State forums now set retail curtailment priorities, as Commissioner David Rosner’s concurrence urged. Quantum gives state commissions a clean test case: a small load with a real recovery cost. Defining protected load by function and recovery time keeps the category narrow as other technologies seek the same treatment (P-1).

💼 Executives. Developers, colocation hosts and hyperscalers share one design answer. Isolate refrigeration and real-time control on a protected circuit, size ride-through to the equipment’s tolerance and keep the GPU campus under ordinary large-load terms (P-1). Hyperscalers placing coupled racks should plan host nodes within a few hundred meters of the processor (S-1).

⚖️ Counsel. Support-system energy data, continuity sizing and recovery tests now document several foreseeable harms. Each can implicate a duty the client already carries: candor in regulatory filings, care in hosting terms and accuracy in sustainability statements. The Foresight-to-Duty list below shows who holds each measurement and who can fund the fix.

📊 Investors. Quantum readiness has option value but no near-term capacity value. A relief disclosure would most likely raise output rather than lower consumption (P-2). Hardware milestones such as S-1 confirm coupling progress, not energy savings. If a site discloses relief, test consumption rebound and contractual capacity release separately (P-2).

Quantum Gaps Become Documented Knowledge Before Any Quantum Rule Exists

MindCast’s Foresight-to-Duty Conversion in AI Governance, Series Introduction (2026) shows how an institution’s own measurement becomes evidence once it keeps the gap it found. The framework applies here because the quantum calculus produces those measurements early: support-inclusive energy data, continuity sizing and recovery exposure.

Judge Learned Hand’s formulation weighs the burden of precaution against the probability and magnitude of harm together. For a quantum island, the energy quantity is modest: an illustrative 1-megawatt system needs about 8.9 megawatt-hours of storage for eight hours.

Energy quantity alone does not establish installation cost, integration feasibility or how the burden compares with expected harm. Documented awareness of an unprotected subsystem therefore counts as relevant evidence, and whether precaution was due requires a case-specific assessment. The party that can fund a correction often differs from the party that verifies it, so the list keeps them apart.

• Refrigeration and control without ride-through (P-1). Host engineering holds the measurement and the host funds the fix. Vendor recovery tests and the tenant verify it.

• Efficiency claim without support-system data (P-1, P-2). The operator’s regulatory team holds the measurement and the operator funds the fix. Commission staff verify it.

• Relief disclosed while consumption holds (P-2). The operator’s sustainability team holds the measurement and the operator funds the fix. Utility metering and investors verify it.

• Coupled racks planned on a vendor schedule (S-1). Hyperscaler infrastructure planning holds the measurement and the hyperscaler funds the fix. Vendor milestones verify it.

Takeaway. Quantum gaps become documented knowledge before any quantum-specific rule exists, so exposure can arrive first through a client’s own files.

Unmeasured Quantum Claims Should Leave Pending Filings First

The sequence runs from what a team controls this month to what a legislature could change next year.

• Now to 30 days: strip unmeasured quantum claims from pending submissions before a commission requests measurement support.

• 30 to 90 days: draft ride-through and recovery clauses before a quantum tenant requests protection.

• 3 to 12 months: separate electrical boundaries in new campus designs before a state opens a large-load rulemaking.

• Beyond 12 months: keep designs that satisfy either regime in case a statute names a facility class as critical load.

Broad Protection Requests, Label Exemptions and Early Credit Carry the Exposure

Entry 1 · 💼 Developers · P-1 (90–100%).

• Exposure: a developer that seeks campus-wide protection around a quantum tenant loses permitting schedule when the forum demands an engineered boundary. Severity: moderate. Probability: high once an instrument appears.

• Mitigating actions: the electrical engineering lead isolates refrigeration and control on a separate circuit before the first 2027 large-load filing. The regulatory affairs lead frames requests by function and recovery time through December 31, 2028. The commercial lead assigns ride-through duty to the host and recovery duty to the vendor before the second hosted installation.

• Residual exposure: a forum may still classify the whole campus by facility label during a pause.

Entry 2 · 🏛️ Policymakers · P-1 (90–100%).

• Exposure: a commission that grants a label-based exemption sets a precedent that other technologies will cite within one or two session calendars. Severity: high. Probability: low.

• Mitigating actions: commission staff publish a functional test for protected load before the 2027 sessions. Legislative staff tie any critical-load designation to demonstrated recovery cost and on-site backup.

• Residual exposure: a statute passed under pro-development pressure can override commission tests.

Entry 3 · 📊 Investors · P-2 (50–60%).

• Exposure: an underwriting case that credits quantum relief before 2031 overstates free capacity by the full credited amount. Severity: high for capacity-constrained valuations. Probability: moderate given a disclosure.

• Mitigating actions: the credit committee grants zero capacity credit to quantum efficiency until a measured disclosure exists. The diligence lead requires complete-system energy data against a tuned classical baseline. The portfolio lead values quantum-inspired software separately from quantum hardware.

• Residual exposure: an early advantage on a narrow workload could favor competitors who credited it sooner.

Entry 4 · 💼 Hyperscalers · S-1 (50–60%).

• Exposure: a hyperscaler that plans coupled quantum racks on the vendor’s announced schedule can strand floor space and host nodes if installation slips. Severity: low to moderate. Probability: moderate.

• Mitigating actions: the infrastructure planning lead reserves convertible space rather than dedicated space through 2027. The procurement lead ties host-node orders to installation milestones.

• Residual exposure: early competitors may secure scarce coupling expertise first.

Nothing in Section IV constitutes legal, investment or fiduciary advice.

Section conclusion. Separation protects against both outcomes of P-1. Separated designs conform if backup duties spread, and they remain approvable if an exemption arrives instead.

V. The First Qualifying Instrument Sets the Template for Quantum Continuity

The first qualifying instrument will set the template for quantum continuity, and its effects travel through institutions, hosts and investors. Two other levers, shared workloads and clean-matched compute, matter more for AI data centers before 2031 than quantum does.

A. State Forums Will Define Protected Load for Small Sensitive Equipment

State forums will define protected load for small sensitive equipment, since FERC treated retail curtailment as state jurisdiction. Given that a qualifying instrument appears, P-1 expects it to define protection by equipment function and backup duty. State-law consequences follow only when a public authority issues that instrument, while a private hosting specification shapes commercial practice alone.

B. Hosts With Engineered Continuity Capture the Early Quantum Tenants

Hosts with engineered continuity capture the early quantum tenants, and they gain bargaining power because they write the first private terms. Storage and backup suppliers sell continuity quality rather than energy volume. Utilities keep curtailment flexibility because the protected island stays small.

Investors who wait lose little. The quantum option carries value for future workloads, while near-term capacity value stays close to zero (P-2).

C. Only a Documented Consumption Decline Counts as Physical Relief

If P-1 fails, the first instrument granted an exemption or critical-load status without a backup duty. The result weakens the predicted preference for subsystem protection, yet the subsystem design remains approvable. Where a state authority issued the instrument, other technologies will cite it.

If P-2 fails, the series identifies which failure occurred: flat output, lower consumption or missing data. Only a documented decline in consumption counts as evidence of physical relief.

Takeaway. The first qualifying instrument sets the template for quantum continuity, and added output determines whether relief reaches the grid.

D. Shared Workloads and Clean Specialized Compute Help Now Far More Than Quantum

Two levers do more than quantum: workload sharing across remote sites and specialized compute matched to clean power. Both help AI data centers now far more than quantum will before 2031.

How much do shared workloads help now? Flexibility has become the currency of approval. Duke University’s Nicholas Institute estimated in February 2025 that U.S. grids could absorb 76 gigawatts of new load if those loads accept curtailment 0.25 percent of the time. At 1 percent, the figure rises to 126 gigawatts.

Shared workloads supply the curtailment flexibility the Duke estimate assumes. Inference, batch jobs and checkpointed training can move between sites, so a campus can shed load while the work continues elsewhere. Large training runs resist splitting across sites, but they can pause between checkpoints.

How much does clean-matched specialized compute help now? Smaller sites matched to their power source carry a lighter and measurable local burden. Burden-based authorization rewards that profile, and the benefit grows where backlash runs strongest. Shifting flexible work toward clean hours and regions also cuts emissions without cutting total demand.

How much will the same levers help quantum-coupled data centers? Shared workloads and clean-matched compute help quantum as a siting architecture rather than as a source of relief. Job-level access lets a few quantum islands serve many remote GPU campuses, so not every campus needs cryogenics or a protected circuit. Firm clean hosts with storage and shared cryogenic plants suit a quantum island’s small and continuous load.

Why keep quantum islands apart from flexible campuses? Workload sharing makes GPU campuses curtailable, while quantum islands cannot pause without recovery costs. Combining them in one campus raises the firm-power intensity of the load that stays, the firmness inversion described in The Grid-Anchored Clean Power Bargain: Powering AI Data Centers Through 2040 (2026). Specialized hubs keep flexible campuses flexible, while real-time decoding stays within a few hundred meters of each processor.

The lever comparisons are assessments rather than Simulation Predictions, and each installment’s simulation will test the one it owns.

Takeaway. Shared workloads and clean specialized compute are today’s approval levers, and quantum inherits them as a siting architecture.

Section conclusion. Quantum’s near-term institutional effect runs through one document, the first continuity instrument, while its siting future runs through levers that already help AI data centers today.

VI. Five Installments Turn the Overview’s Mechanisms Into Site-Level Forecasts

Each installment owns a distinct set of Simulation Predictions and carries its own simulation. The overview’s releases do not repeat in the installments.

• Power grid: does hybrid computing cut demand, reshape it or raise the value of continuity? The answer informs capacity reservations and backup design.

• Cross-border compute: which work stays local and which moves across continents? The answer informs where to place quantum islands.

• Clean technology market: what can clean energy developers sell into hybrid computing? The answer informs product and hosting strategy.

• Environmental balance: does quantum avoid more burden than its complete infrastructure creates? The answer informs environmental disclosure and review.

• Data center approval: when does quantum make a project easier or harder to authorize? The answer informs approval strategy.

VI.A Power Grid Relief Depends on Support Overhead and Output Growth

Leading insight. Consumption falls when displaced classical energy exceeds supporting overhead and added output stays below the savings threshold. Reserved capacity needs its own test, because a site can consume less while contracting the same capacity. The installment tests the relief hypotheses of How Quantum Computing Overcomes AI Data Center Bottlenecks (2025) against support overhead and output growth.

What the simulation will test. The power grid installment will model operator, host and utility choices across measured support overhead by platform. The installment will test whether a site’s reserved capacity falls even when energy per result does. Three comparisons structure the test.

Adding quantum to an existing AI facility adds cooling, controls and supporting classical load. Replacing a classical workload cuts consumption only if the complete quantum workflow uses less energy for an equivalent result. Using the savings for more output can lower energy per result while total consumption holds or rises.

What it cannot establish. Behavioral simulation cannot establish experimental energy advantage. Uncertain technical performance enters as competing inputs.

Takeaway. The grid installment turns the energy accounting into site-level Simulation Predictions on consumption, reserved capacity and peak.

VI.B Real-Time Control Stays Local While Job-Level Access Crosses Borders

Leading insight. Globally distributed access can connect locally concentrated computing islands. Real-time decoding stays within one building, and linked modules work within a campus or metro area. Job-level access crosses continents at round trips of tens to hundreds of milliseconds.

What the simulation will test. The cross-border installment will model which allied jurisdictions win the computing islands under export controls, data-localization law and sovereign-cloud rules.

The U.S. Bureau of Industry and Security’s September 2024 quantum controls govern hardware exports, and remote access remains the open question. Quantum Computing Sovereignty (2025) supplies the premise that integration intellectual property decides access.

What it cannot establish. The cross-border installment cannot forecast distributed quantum computation across continents, which would require entanglement distribution beyond current capability.

Takeaway. Location follows export status, cryogenic supply and talent more than power.

VI.C Clean Technology Sells Continuity and Cooling Into Hybrid Computing

Leading insight. The computing islands need continuity, cooling and recovery services more than energy volume. Clean firm power fits a quantum island but cannot drive procurement. A firm clean host can bundle those services with an allied jurisdiction. The Clean Compute Match (2026) supplies the pairing of firm clean power with anchored compute that the installment extends.

What the simulation will test. The clean technology installment will add firm clean developers, storage suppliers and cryogenic service providers as Cognitive Digital Twins. The installment will also track quantum-assisted materials discovery, such as the Department of Energy’s $37 million Quantum Computing for Computational Chemistry (QC3)program, as a separate opportunity.

What it cannot establish. Discovery benefits require experimental validation and manufacturing. The installment will treat them as staged gates without a fixed arrival date.

Takeaway. The clean technology market for quantum sells service quality, not kilowatt-hours.

VI.D Environmental Burden Moves With Quantum Architecture

Leading insight. The environmental installment assesses the total burden of the computing islands, and exposure differs by architecture. Superconducting and silicon spin systems depend on dilution refrigeration and helium-3, while trapped-ion and neutral-atom systems face different constraints. Prospective research such as Oak Ridge National Laboratory’s study of quantum computing futures concerns scaled superconducting systems in the 2030s and 2040s.

What the simulation will test. The environmental installment will compare electricity, manufacturing and water burdens against equivalent useful work, then add cryogens and rebound. The installment will model the helium-3 supply chain, including forward contracts for lunar helium-3 that depend on unproven extraction.

What it cannot establish. Prospective discovery benefits cannot offset present local burdens. The installment will keep them on separate scenario clocks.

Takeaway. Environmental exposure moves with architecture, so no single quantum footprint exists.

VI.E Approval Turns on Measured Burden and Claim Credibility

Leading insight. Quantum can change approval prospects through measured changes in project burden and through the credibility of the developer’s claims. Unmeasured offset claims lose credibility in forums that demand measurement. Engineered quantum islands offer authorities what they already require. Burden-Based Authorization for AI Data Centers, Modular Clean Compute, and the Backlash Gap Between Grid Tariffs and State Law (2026) supplies the measured-burden test the installment applies.

What the simulation will test. The approval installment will compare three project types against two developer postures. The project types are additions to existing facilities, dedicated quantum installations and new AI campuses claiming quantum benefits. The postures are claimed benefit and measured burden, which yields six approval cells. County and community Cognitive Digital Twins join the state authorities.

What it cannot establish. The approval installment cannot forecast the outcome of a specific pending permit without that project’s filings.

Takeaway. The finale answers whether quantum changes a project’s actual burden enough to change permission to build.

Section conclusion. The installments move from physics to permission: grid and environment test what quantum changes, cross-border and clean technology test where it locates and the finale tests whether authorities let it build.

Hosting Specifications and Relief Disclosures Resolve the Forecasts

Seven signals resolve or move the overview’s forecasts, and most arrive through hosting specifications, commission dockets and operator disclosures.

• Activates P-1: quantum hosting specifications and service agreements, reviewed quarterly in colocation operator publications.

• Shifts P-1 route weights: state large-load rulemakings and curtailment designations, reviewed monthly in commission dockets.

• Breaks P-1: critical-load legislation, reviewed during legislative sessions.

• Activates P-2: measured relief disclosures from operators and vendors, reviewed quarterly.

• Resolves P-2: output and metered demand at disclosing sites, in operator disclosures and interconnection filings.

• Resolves S-1: Superion installation status in IonQ and NVIDIA releases, reviewed quarterly.

• Releases watch conditions: the Texas queue audit and permit decisions, reviewed monthly.

Section conclusion. P-1 and S-1 resolve against public documents by 2028, while P-2 activates only if a site discloses measured relief.

Conclusion: Developers Who Measure Quantum’s Burden Keep It From Deepening the Controversy

Quantum lowers AI data center consumption when displaced classical energy exceeds supporting overhead and added output stays below the savings threshold. Through 2030, the simulation finds hybrid computing more likely to shift constraints than relieve them. Continuity moves onto the grid agenda, cryogenics and materials move onto the environmental agenda and export status moves onto the siting agenda. Quantum will therefore leave the controversy around AI data centers unchanged or deepen it, depending on whether developers measure its burden or promise its offsets.

Globally distributed access will connect locally concentrated computing islands. Stakeholders should act now on what they can measure: support-system energy, protected subsystems and the credibility of their own claims. They can wait to learn whether useful advantage arrives on DARPA’s schedule.

The decisive signal is the text of the first qualifying instrument addressing quantum continuity. If it grants an exemption without a backup duty, P-1 fails. The predicted preference for subsystem protection then weakens, while the subsystem design remains approvable.

A Campus’s Quantum Hosting Decision Depends on Confidential Inputs

Public analysis cannot tell a specific developer whether its campus should host quantum equipment. The answer depends on the site’s contracts, its forum and its customers’ workloads. Site inputs stay confidential.

MindCast builds Cognitive Digital Twins of the developer, its utility, its state authority and its prospective quantum tenants. The engagement ingests the organization’s own decision history and reruns the continuity contest and energy accounting against its actual options.

Commissioned scope adds complete probability distributions, P10/P50/P90 ranges and organization-specific calibration. The scope also adds complete scenario trees, decision thresholds, trigger dashboards and red-team tests of counterparty responses.

The bounded next step is a scoped site diagnostic: one campus, one forum and one quantum hosting decision. Contact mcai@mindcast-ai.com or visit www.mindcast-ai-simulation.com to arrange it.

Appendix A. Selected MindCast Works

The series

• Which Funded Solutions Keep Quantum-Coupled AI Data Centers Approvable? (October 2026). Tests 31 measures against the burdens the overview names and assesses which keep quantum sites approvable.

Frameworks and methods

• The Dynamic Predictive Game Theory Collection (July 2026). Supplies the changing-game frame used to model curtailment treatment in state forums after FERC’s October 9 order.

• Field-Geometry Reasoning: A Unifying Framework for Structural Explanation in Law, Economics and Artificial Intelligence (January 2026). Supplies the constraint lens that separates workload, coupling, continuity and materials constraints.

• The Authorization Market: Standardized Bargaining, Rationed Power, and the Competition to Build America’s AI Infrastructure (August 2026). Supplies the finding that connected power is not firm power, which frames the continuity contest.

• Burden-Based Authorization for AI Data Centers, Modular Clean Compute, and the Backlash Gap Between Grid Tariffs and State Law (September 2026). Supplies measured burden in place of the facility label, the test the approval installment applies.

• Foresight-to-Duty Conversion in AI Governance, Series Introduction (October 2026). Supplies the conversion of measured gaps into evidence of awareness, the frame for the Foresight-to-Duty exposure list.

Prior quantum analysis

• Quantum Computing and the Future of Economics (May 2025). Frames quantum as a change in how institutions model decisions under uncertainty.

• How Quantum Computing Overcomes AI Data Center Bottlenecks (October 2025). Supplies the relief hypotheses that the energy accounting now tests against support overhead and output growth.

• NVIDIA NVQLink Validation (October 2025). Supplies the microsecond coupling analysis behind the co-location limit.

• Quantum Computing Sovereignty (October 2025). Finds that integration intellectual property decides access, a premise of the cross-border installment.

• The AI-Quantum Closed-Open-Hybrid Architecture Problem (March 2026). Grounds the tightly coupled tier that cannot be distributed.

• AI × Quantum × Blockchain Convergence, A Three-Clock Framework for Capital Allocation, Verification, and Fault-Tolerant Optionality (April 2026). Separates the clocks that keep discovery benefits apart from near-term AI load.

Related AI data center analysis

• The Grid-Anchored Clean Power Bargain: Powering AI Data Centers Through 2040 (September 2026). Supplies the firmness inversion: moving flexible compute toward energy raises the firm-power intensity of the load that stays.

• The Clean Compute Match: Which Clean Power Pairs With Which AI Compute, And What Stays On Fossil(September 2026). Pairs firm clean power with anchored compute, the hosting route the clean technology installment tests.

• AI Data Center Veto Points Show Where Projects Get Stopped and How Authorization Risk Moves (September 2026). Maps the approval stages where quantum claims would enter.

Appendix B. External Sources

• DARPA, “Four More Teams Enter Quantum Benchmarking Initiative Final Stage“ (October 7, 2026).

• FERC, Order on PJM Interconnection Large Load Proposal, Docket ER26-3515-000 (October 9, 2026).

• Bracewell, “Governor Abbott Orders TCEQ to Pause All Permit Issuance for Data Center Projects“ (September 23, 2026).

• Equinix, “Diraq and Equinix to Deploy Australia’s First Quantum Computer Inside a Data Centre“ (September 1, 2026).

• IonQ, “IonQ to Advance Quantum Supercomputing by Bringing First QPU to NVIDIA Accelerated Quantum Research Center“ (September 23, 2026).

• NVIDIA, “NVIDIA NVQLink Architecture Integrates Accelerated Computing with Quantum Processors“ (2025).

• Quandela, NVQLink demonstrator announcement (October 13, 2026).

• Carrasco-Codina et al., “Energy Efficiency of Quantum Computers,” arXiv:2605.15090v2 (2026).

• Oak Ridge National Laboratory, “Uncertain Quantum Computing Futures and Potential Energy and Physical Resource Implications.”

• U.S. Bureau of Industry and Security, quantum computing controls (September 2024).

• GeekWire, Interlune helium-3 agreements with DOE and Maybell Quantum (May 2025).

• Duke University Nicholas Institute, “Rethinking Load Growth“ (February 2025).

• Utility Dive, “FERC Rejects Core of PJM Large Load Interconnection Plan“ (October 2026), reporting Commissioner Rosner’s concurrence.

• ARPA-E, Quantum Computing for Computational Chemistry (QC3) program announcement.

Foresight-to-Duty Conversion Series Introduction: Meta Foresight-to-Duty Conversion Across Privacy, Child Safety and Scam Ads · Meta and Bank of America Foresight-to-Duty Conversion: Revenue-Funded Versus Cost-Funded Deferral · Consent Order Remedies at Meta, Boeing, Starling and Wells Fargo: Verified Correction Versus Recurrence

Related Works: Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much · The Foresight-to-Duty Conversion — Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found

Foundational Works: The Duty to Foresee, AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care · AI Governance Equilibrium · Decision Modeling and Foresight Simulation

Executive Summary

Bank of America and Meta kept gaps their own systems measured for opposite economic reasons. Bank of America avoided an operating expense. Meta allegedly preserved revenue and earned more from the fraud signal itself. The paper tests one hypothesis: the source that funds a deferral decides who controls it, and the controller decides how long it lasts.

Thesis: Both institutions met the same deferral inequality: the expected sanction looked smaller than the cost of correction. Bank of America’s 2022 test showed an unconstrained threshold would surface 85.4% of sampled suspicious activity against 20.2% under the existing rule, and management kept the rule. Meta’s internal documents estimated 15 billion higher-risk scam ads a day and 10.1% of 2024 revenue from scam and banned-goods ads.

Bank of America admitted willfulness in the October 13, 2026 consent order. Meta disputes the allegations against it. The admitted case anchors the mechanism the alleged case tests.

What distinguishes the MindCast approach: Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much (2026) identified capacity-calibrated detection: a bank sized what its monitor could see to the investigators it chose to fund. The Foresight-to-Duty Conversion (2026) formalized the deferral inequality that sustains such a gap. The present paper adds the variable both held fixed: the source that funds the deferral, set against the authority that approves it. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through the MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation (MP CDT FS) system, with Behavioral Economics supplying the decision rules and Game Theory the payoff structure.

What the paper does: The paper states the shared inequality and the two funding sources, then places each stall with its controller. It sets aggregation against per-violation scaling as routes to correction and shows how activity metrics hid both gaps. It examines how automation changes review capacity, frames the comparative cases and closes with three Simulation Predictions and a Risk Mitigation Layer.

Simulation Prediction highlights: The forecast set carries three MindCast Foresight Simulation Predictions: one primary and two secondary. The lead entry forecasts that Bank of America’s regulators publicly report effectiveness-tested remediation before Meta publicly discloses a verified change to flagged-advertiser treatment and do so by December 31, 2028 (47-60%). The forecast supports an authority-based reading and does not discriminate the independent effect of funding source.

🏛️ Policymakers: Per-violation statutes can reach revenue-funded deferrals that fixed fines miss. New Mexico’s $25 billion privacy judgment shows the scale such a statute can produce in one state.

💼 Executives: Risk and compliance leaders: the approver’s budget target, not the finding, decides when a correction funds. Revenue owners: a fix authorized against your own results stalls until approval moves outside your line. Boards: a glide path is a scheduled deferral, and its approval is a dated decision.

🤖 AI deployment leaders: Automation can move the deferral from detection to escalation. The budget decision on escalation capacity then becomes the file.

⚖️ Counsel: A surcharge on flagged advertisers turns a fraud signal into an income line. The surcharge adds evidence of incentive to evidence of knowledge.

📊 Investors: A cost-funded deferral ends when the bank funds review capacity. A revenue-funded deferral ends with a revenue reset, so disclosed practice changes carry more information than enforcement totals.

Across audiences the implication is the same. The approver’s position, not the size of the gap, sets when a correction funds.

I Two Institutions Met One Inequality From Opposite Economics

The deferral inequality from the companion paper holds when the discounted expected sanction falls below the per-period cost of correction, which in turn falls below the per-period harm the correction would prevent. Bank of America satisfied it by avoiding an operating expense. Meta allegedly satisfied it by preserving contribution.

The inequality says nothing about who decides. Two institutions can meet the same condition with different controllers, and the controller’s incentive decides how the inequality resolves once a sanction arrives. Funding source is the paper’s candidate for what places the controller.

Four quantities enter the comparison over the same period: the cost of correction, the contribution retained by deferral, the expense avoided by deferral and the expected sanction. A deferral can retain contribution and avoid expense at once, so the paper does not assign each institution wholly to one side.

The October 2026 calendar supplies a useful comparison. FinCEN’s order against Bank of America and Rhode Island’s complaint against Meta arrived nine days apart, so both deferrals face enforcement from nearly the same start date. The instruments still differ: a binding bank order with admitted willfulness against a newly filed platform complaint.

The table sets the two deferrals side by side.

Takeaway: The inequality is shared, and funding source is the variable under test rather than the answer.

II Bank of America Deferred an Operating Expense

The Bank of America paper holds the chronology, and the present paper cites it. The 2022 test compared 85.4% against 20.2%. Management kept the existing threshold on investigation-volume grounds, and a 2024 estimate put the staffing an unconstrained threshold would require at roughly 750 people against roughly 16,000 monthly investigations.

A cost center held the correction. The budget owner would have absorbed the headcount expense, and that owner’s performance metric was spend against plan. A cost center benefits from deferral when its target rewards flat expense, so the behavioral mechanism runs through a target rather than through revenue.

Separating the bank’s compliance function from its funding authority shows where the gap sat. The compliance function measured the gap in 2022 and could size the fix, but it held no budget authority over headcount. The funding authority held the budget, and the order’s independent review now sets the requirement it must meet.

Takeaway: The cost-funded deferral sat with a funding authority whose budget target rewarded it, apart from the function that found the gap.

III Meta Preserved Revenue and Allegedly Monetized the Signal

Meta’s reported internal documents set a glide path for scam and banned-goods advertising from 10.1% of revenue in 2024 to 7.3% by the end of 2025 and 6% by 2026. A glide path is a planned deferral: the firm scheduled the correction over several years and kept the contribution in the interval. Revenue guardrails limited enforcement teams from acting against advertisers whose removal would cost more than 0.15% of total revenue.

Reuters reported a second step. Flagged advertisers paid higher rates through penalty bids rather than losing access, which converted the detection signal into an income line. The relevant quantity is the incremental contribution Meta retained by not blocking, net of what blocking would have cost, which the firm’s own accounts can date.

Takeaway: The revenue-funded deferral earned contribution rather than savings, and the controller’s results depended on it directly.

IV Who Controls the Stall Decides How Long It Lasts

Both institutions wrote their deferral into an approval rule. Meta’s enforcement teams could not act against an advertiser whose removal would cost more than 0.15% of total revenue, about $135 million in the first half of 2025, without clearing the revenue line. Bank of America’s threshold stayed where investigation volume could absorb it, which tied detection to the staffing a budget owner chose to fund.

Behavioral economics reads the two rules as one decision with different salient metrics. At Meta the approver sees lost revenue on every removal above the guardrail. At Bank of America the approver sees added headcount on every threshold change. Neither approver sees the harm the gap allows, so each rule holds until the harm acquires a number on the approver’s own line.

Game theory locates the break point in who signs. A profit-center approver pays the correction out of its own results, so its rule holds until a sanction exceeds the contribution at stake. A cost-center approver pays out of a budget line the firm can reset in one cycle, so its rule can break once an enforcer converts the gap into a mandated requirement.

The guardrail also supplies an observable. A disclosed change in who approves removals above the guardrail would show authority moving. A raised threshold with the same approver would show the rule loosening while authority stays put.

The Simulation Predictions support an authority-based forecast but do not discriminate the independent effect of funding source. A cost center and a profit center may behave alike when one actor holds both the budget target and the approval. The two may diverge when approval sits outside the line that bears the cost.

Takeaway: Both institutions encoded deferral as an approval threshold, and the threshold’s owner sets how long the gap lasts.

V Two Routes Break a Revenue-Funded Deferral

Game theory supplies the first route. Sanctions across channels add until the sum reaches the contribution line, which requires several enforcers and several years. Installment I treats that aggregation.

New Mexico supplies the second route. A per-violation statute multiplies a fixed penalty by affected users, and the user base that funds the revenue also funds the sanction. The October 2026 judgment reached $25 billion on privacy conduct after the state sought $35 billion to $45 billion, while Meta’s documents separately estimated roughly $16 billion a year in revenue from scam and banned-goods ads.

The two figures cover different conduct and different periods. Together they show the scale per-violation exposure can reach in one state, not that the privacy judgment crossed the scam-advertising break point. No Simulation Prediction carries a band on the per-violation route, because the public file does not yet show how an appellate court will treat the counting.

The New Mexico Court of Appeals decides how far the second route extends. An affirmance above $10 billion would make per-violation exposure concrete under New Mexico’s statute and a reference point for states with comparable provisions. A sharp reduction or vacatur would leave multistate aggregation as the main route in New Mexico.

Takeaway: The user base that funds the deferral can also fund the sanction where a per-violation statute applies, and the New Mexico appeal sets the reach of that route.

VI Activity Metrics Hid Both Gaps

Bank of America reported steady case counts driven by referrals. The count measured investigations opened, not suspicious activity detected, and Section II.D of the Bank of America paper shows how a stable activity metric masked a detection rate of roughly one in five.

Meta reported more than 134 million scam ads removed in 2025 and a 58% decline in user reports over 18 months. Both figures measure activity and response. Neither measures prevalence, which the internal estimate of 15 billion higher-risk ads a day did measure.

Activity metrics persist for a behavioral reason. Activity counts rise with effort and fall only when effort falls, so a team that reports them can show progress while the underlying gap widens. Prevalence and detection rates move with the gap itself, which gives an approver the least reason to publish them.

Automation sharpens the asymmetry. Automated review can multiply removal counts at near-zero cost while prevalence stays unmeasured in public. A board that receives activity metrics alone cannot tell whether a correction is working.

Takeaway: Both institutions reported activity while effectiveness stayed internal, so prevalence and detection rates are the metrics a board should demand, and Installment I forecasts whether Meta publishes one.

VII Admitted Willfulness and Alleged Deception Run Through Different Legal Channels

required Bank of America to detect and report, so the 2022 test went directly to the willfulness the consent order admits.

Platforms carry no equivalent statutory duty to detect scam advertising. Rhode Island’s claims run through representation: Meta’s public anti-scam statements against its internal estimates. Section V.F of the Bank of America paper treats the opinion doctrine that governs such claims.

Meta’s likely immunity defense under the federal platform statute reaches the hosting of third-party ads. It does not reach the firm’s own statements or its own alleged role in creating the ads.

The activity metrics from Section VI become the statements at issue. Meta’s public figures on removals and declining user reports sit beside internal estimates of 15 billion higher-risk ads a day, and Rhode Island’s representation claim tests the distance between them. The metric that hid the gap becomes the statement that creates the exposure.

The difference in legal posture also changes what each institution can still contest. Bank of America admitted willfulness, so its remaining choices concern how fast and how visibly it corrects. Meta disputes the allegations, so its choices still include litigating the meaning of its own documents, which lengthens the period before any correction becomes visible.

Takeaway: The representation channel reaches what the platform statute does not, and it turns the activity metrics that hid the gap into the statements a court will test.

VIII Automation Changes the Cost of Review Capacity

Section V.E of the Bank of America paper located the deferral in the cost of review capacity. An unconstrained threshold would have produced findings the funded investigators could not clear, and the bank’s own estimate put the required staffing at roughly 750 people. Automation changes every term in that calculation.

Automated detection raises finding volume at near-zero marginal cost. Three outcomes are possible. Automated triage can clear the backlog and end the deferral. Triage can lower the cost of clearing a finding while raising the false-positive share that reaches a human, which shifts the bottleneck to escalation.

Findings can also expand faster than correction capacity, which enlarges the deferral. Escalation capacity, the number of findings a decision-maker can act on in a period, can expand with investment. Whether it does is a budget decision the same approver makes, and the three outcomes remain competing hypotheses.

Meta’s reported configuration resembles the second outcome. AI ad-creation tools raise the volume of content integrity systems must assess, and the 95% ban threshold functions as an escalation threshold set within revenue guardrails. The public file does not show which outcome holds once capacity is funded at either institution.

Agentic deployments inherit both structures. A pre-deployment evaluation that surfaces more findings than the review queue can clear produces dated knowledge at machine speed, and each scale-up gate renews the deferral. The Foresight-to-Duty companion’s P-2 forecasts whether an enforcer reads such an evaluation as knowledge evidence, and the present paper supplies the capacity economics behind that reading.

Takeaway: Automation can clear, shift or enlarge a deferral, and the budget owner who funds escalation capacity decides which.

IX Comparative Cases Frame the Funding-Source Hypothesis

A comparative test of funding source would measure deferral duration from dated internal measurement to verified closure across revenue-funded and cost-funded cases. The candidate cases appear in Appendix C with the dates the public file supports. Every cost-funded case remains open, and several measurement dates are not public, so no duration comparison is possible yet.

Two confounds would shape any such test. Enforcement structure differs, because banking operates under a statutory detection duty and platforms do not. Industry differs with it, and the two revenue-side cases reached different instruments: Amazon a negotiated order, Meta a contested complaint.

The hypothesis has a clear falsifier. If closed cases show no duration difference between funding sources once authority placement is known, funding source adds nothing independent. A result favoring authority placement over funding source would be a finding in its own right.

Takeaway: The comparative cases define the test for funding source, and the test waits on cases that have not yet closed.

X Three Explanations Compete, and the Forecast Rests on Authority

Three variables could set how long a measured gap stays open: who funds the fix, who approves it and how large the expected sanction is. Each appears below as a hypothesis with its falsifier.

Hypothesis A: funding source sets duration: Revenue-funded deferrals last longer than cost-funded deferrals once enforcement arrives. Falsifier: closed cases show equal durations once authority placement is known.

Hypothesis B: authority placement sets duration: Deferrals end faster when approval sits outside the line that bears the cost, whatever the funding source. Falsifier: relocated approval fails to shorten deferral in closed cases.

Hypothesis C: expected sanction sets duration: Deferrals end when the expected sanction exceeds retained contribution or avoided expense, whatever the controller. Falsifier: a deferral ends while the expected sanction stays below the contribution line.

The Simulation Predictions forecast public evidence of correction under Hypothesis B. Two per-violation hypotheses remain open without bands: that a new state per-violation judgment exceeds the reported annual revenue from the conduct at issue, and that Meta changes a disclosed scam-ad practice within two quarters of an affirmance in New Mexico.

XI MindCast Simulation Predictions

The forecast set carries three Simulation Predictions: one primary (P) and two secondary (S). Verification sources and partial-outcome rules for each entry appear in Appendix B.

P-1 · Bank of America’s regulators publicly report effectiveness-tested remediation before Meta publicly discloses a verified change to flagged-advertiser treatment and do so by December 31, 2028 (47-60%)
The entry forecasts the order of public evidence, not which institution corrected first in operation. Falsifier: Meta’s verified public disclosure arrives first, or no qualifying bank report appears by the deadline.

S-1 · Meta announces it will block rather than surcharge advertisers its systems flag as likely scammers by December 31, 2027 (42-53%)
A change disclosed only in a court or regulatory filing counts. Falsifier: no public disclosure by the deadline.

S-2 · A securities complaint ties Meta’s anti-scam statements to its internal estimates and survives a motion to dismiss on those statements by December 31, 2028 (22-33%)
Survival on unrelated statements does not count. Falsifier: dismissal, or no complaint by the deadline.

The three entries forecast public evidence rather than private correction. P-1 tests whether the bank’s regulators report verified remediation first. S-1 and S-2 test whether Meta’s flagged-advertiser practice and its public statements change under pressure.

Takeaway: The forecasts read the order of public evidence, which is what outside observers can verify.

XII Risk Mitigation Layer

Exposure appears in a unit each stakeholder controls, with severity and probability stated separately.

Platform revenue leadership · P-1 (47-60%), S-1 (42-53%): Severity: high. Exposure: each quarter of deferral adds dated contribution to the discovery file.

• Action 1: move approval of flagged-advertiser treatment to an owner outside the revenue line and document the transfer, owner the chief business officer, before the Rhode Island motion-to-dismiss ruling.

• Action 2: replace penalty bids with blocking at the current detection threshold and disclose the change, owner the chief business officer, within two quarters.

• Residual: the prior surcharge period remains dated knowledge.

Bank compliance and funding leadership · P-1 (47-60%): Severity: high. Exposure: a budget target that rewards flat spend can reproduce the 2022 decision under the order.

• Action 1: size the review capacity the independent review requires and escalate the request in writing, owner the chief compliance officer, before the first supervisory follow-up.

• Action 2: fund the requested capacity and document the approval, owner the chief financial officer, same deadline.

• Residual: an effectiveness-tested follow-up shows operational correction and not durability. The second action also reduces exposure in the Bank of America paper, and credit is claimed once.

AI deployment owners · P-1 (47-60%): Severity: moderate. Exposure: an evaluation whose findings exceed escalation capacity can reproduce the Bank of America threshold decision at each scale-up gate.

• Action 1: size escalation capacity to the finding volume the evaluation produced and document the sizing decision with its approver, owner the deployment owner, before each gate.

• Action 2: route the funding decision to an approver outside the product line, owner the AI governance lead, same deadline.

• Residual: a funded queue clears findings and leaves the pre-funding period as dated knowledge.

Disclosure counsel · S-2 (22-33%): Severity: high. Exposure: each public anti-scam statement measured against the internal prevalence estimate is a candidate misstatement.

• Action 1: reconcile each public statement to the internal figure it rests on and retire statements that fail, owner disclosure counsel, before the next quarterly filing.

• Action 2: convert removal-count statements to prevalence-rate statements with method, owner disclosure counsel, same deadline.

• Residual: the reconciliation documents the gap for the period before it.

Institutional investors · S-2 (22-33%): Severity: moderate. Exposure: a surviving securities complaint opens discovery into the internal estimates behind Meta’s public statements.

• Action 1: model securities-claim exposure tied to the anti-scam statements, owner the governance analyst, before the next annual report.

• Action 2: submit a disclosure question on how Meta measures scam-ad prevalence, owner the stewardship lead, before the annual meeting.

• Residual: discovery outcomes in Rhode Island can change the exposure faster than any model updates.

The actions share one direction. Each separates the decision to fund a correction from the line that bears its cost, and each leaves the pre-correction period on file.

Takeaway: Moving the funding decision is the lever that shortens deferral.

Conclusion

One deferral inequality held at two institutions from opposite economics. Bank of America’s gap sat with a cost center, and a consent order with a $1.2 billion penalty now requires it to fund the correction. Meta’s alleged gap sits with a profit center the practice pays, so the break requires a sanction that reaches the contribution line or an approver outside the line.

Funding source remains a candidate predictor. The comparative cases define how to test it, and the test waits on cases that have not closed. Authority placement carries the forecast in the meantime.

P-1 forecasts the order of public evidence: Bank of America’s regulators publicly report effectiveness-tested remediation before Meta publicly discloses a verified change and do so by December 31, 2028 (47-60%). A hit supports the authority-based reading without measuring which institution corrected first in operation.

What to Watch

Checkpoints fall on January 21, 2027 and on October 23 of 2027 and 2028. The dominant fork is the basis of the first Bank of America supervisory follow-up.

Each signal maps to an entry. The supervisory follow-up and any Meta disclosure on flagged-advertiser treatment resolve P-1. A Meta announcement on penalty bids resolves S-1. The first securities complaint citing the internal estimates opens S-2.

Working With MindCast

Published analysis can set two institutions’ deferrals side by side. It cannot tell a specific institution what its own deferral is worth, who must approve each fix or how long its gap will stay open under its current approval structure. The answers sit in the institution’s budget files, approval chains and testing history, as the series introduction sets out.

MindCast builds Cognitive Digital Twins (CDTs) of an institution’s measurement, authority and verification system. It ingests the testing history and reruns the enforcement environment against the institution’s actual options.

For the question this installment asks, the engagement is a deferral review for risk and finance leaders, funding authorities and business owners. The review compares the expected return to deferral against the cost of each fix, using the institution’s own numbers rather than public estimates. It names the approver whose position must change for each correction to fund.

AI deployment leaders face the same structure at machine speed. A review-capacity simulation tests whether the human review queue can absorb what pre-deployment evaluation surfaces, and it runs before the first agentic precedent lands. The output sizes escalation capacity to the finding volume the evaluation produces.

Commissioned work delivers three outputs. The first is a map of open findings by stage of conversion. The second is a set of exposure scenarios with low, median and high cases. The third is a set of monitoring thresholds keyed to this installment’s Simulation Predictions: the basis of the next supervisory follow-up (P-1), a change in flagged-advertiser treatment (S-1) and the first securities complaint (S-2).

MindCast takes one bounded next step at a time, and no engagement proceeds without a named decision to resolve. Contact mcai@mindcast-ai.com or visit www.mindcast-ai-simulation.com to schedule the diagnostic session.

Takeaway: The installment compares two deferrals from the outside, and a deferral review measures the deferral inside one institution.

Sources

The works below support the facts and frameworks cited above. MindCast works come first, then primary sources, then secondary reporting.

MindCast Works

• Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much(2026). Supplies capacity-calibrated detection, activity metrics, review capacity and the opinion doctrine.

• The Foresight-to-Duty Conversion: Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found (2026). Supplies the deferral inequality and the agent-deployer forecast.

Primary Sources

• Financial Crimes Enforcement Network, release on the Bank of America consent order, October 13, 2026 ($1.2 billion; Office of the Comptroller of the Currency civil money penalty of $750 million; Office of Foreign Assets Control resolution of $11.6 million).

• Bank Secrecy Act, 31 U.S.C. § 5311 et seq.

• 47 U.S.C. § 230.

Secondary Reporting

• State of New Mexico v. Meta Platforms, Inc., judgment of October 21, 2026, as reported by ABC News.

• State of Rhode Island v. Meta Platforms, Inc., complaint filed October 22, 2026, as reported by PYMNTS.

• Reuters, special report on Meta’s internal scam-advertising documents, November 6, 2025, as summarized by MediaNama.

Foresight-to-Duty Conversion Series Introduction: Meta Foresight-to-Duty Conversion Across Privacy, Child Safety and Scam Ads · Meta and Bank of America Foresight-to-Duty Conversion: Revenue-Funded Versus Cost-Funded Deferral · Consent Order Remedies at Meta, Boeing, Starling and Wells Fargo: Verified Correction Versus Recurrence

Related Works: Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much · The Foresight-to-Duty Conversion — Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found

Foundational Works: The Duty to Foresee, AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care · AI Governance Equilibrium · Decision Modeling and Foresight Simulation

Executive Summary

Separate fines do not diversify governance risk when one decision architecture produces the failures they punish. Meta now faces that test in three legal channels. A New Mexico jury returned a $375 million child-safety verdict in March 2026. A New Mexico judge ordered a $25 billion privacy penalty on October 21, and Rhode Island sued over scam advertising on October 22.

Thesis: Several legally distinct failures at one firm can share one decision architecture. The architecture has three parts: the actor with authority to fund a correction, the metric that actor’s budget tracks and the loop that verifies closure. Meta’s own files dated its awareness in each channel, while the public file names the corrective recipient in only two.

What distinguishes the MindCast approach: The Foresight-to-Duty Conversion: Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found (2026) traced how one measured gap becomes evidence once an institution keeps it. The present paper applies those stages across channels at one firm and asks whether one authority sat behind each stall. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through the MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation (MP CDT FS) system. Behavioral Economics supplies the decision rules, Game Theory supplies the payoff structure, and predictive simulations emerge from the combination.

What the paper does: The paper defines decision architecture and its evidence standard, then dates Meta’s awareness in each channel. It builds an authority map, adds two configurations prior models omit and tests the pattern against two comparators. It forecasts parallel state filings and closes with six Simulation Predictions and a Risk Mitigation Layer.

Simulation Prediction highlights: The forecast set carries six MindCast Foresight Simulation Predictions: two primary and four secondary. The lead entry forecasts that a second state attorney general sues Meta over scam advertising by December 31, 2027 (67-79%). Whether one approver sat behind deferrals in several channels remains an open hypothesis without a band.

🏛️ Policymakers and state enforcers: An order that names who authorized a deferral becomes evidence for the next enforcer in another channel. A file that names no approver leaves each channel to start over.

💼 Executives: Risk and internal audit leaders: an open finding without a dated receipt by a funding authority proves awareness but not authority. Revenue owners: an order in one channel supplies notice in another only where the relevant finding reached a named recipient. Boards: an assessor finding whose recipient is unknown is the gap a plaintiff will fill.

🤖 AI deployment leaders: An evaluation that surfaces more than the review queue can clear produces dated knowledge. The owner who decides what the queue absorbs holds the exposure.

⚖️ Counsel: An enterprise-level claim stands only where one body approved deferrals in several channels. Approval maps therefore decide the scope of liability.

📊 Investors: Meta’s legal exposure behaves as one correlated position rather than three independent matters. A per-violation statute in one state produced a judgment larger than the multistate resolution of the same conduct.

Across audiences the implication is the same. The question that decides Meta’s exposure in each channel is who received the finding and what that recipient could fund.

I One Decision Architecture Can Sit Behind Three Regulatory Channels

Meta’s 2026 enforcement calendar turned three legal channels into one balance-sheet question. Judgments and resolution commitments reached up to roughly $42 billion before Rhode Island’s scam-advertising claim was counted: $25 billion in New Mexico on privacy, $375 million on child safety and up to $17.1 billion in the multistate resolution. Investors and boards see three provisions. The paper asks whether they reflect three decisions or one.

The unit of analysis is therefore the decision architecture, not the violation. A violation names conduct a statute reaches. A decision architecture names the actor who could fund the correction, the metric that actor’s budget tracks and the loop that confirms closure.

The answer changes how the exposure behaves. If one architecture governs all three channels, each enforcer inherits the others’ evidence and the provisions move together. If each channel has its own approver, each matter resolves on its own file.

One evidence standard decides between the two. A claim that Meta was aware of a deficiency requires a dated internal document (Level 2a). A claim that a corrective authority received it requires a named recipient (Level 2b), and a claim of one architecture requires one body that received findings and declined correction in more than one channel (Level 3).

Two kinds of evidence fall short of Level 3. Documents moving between proceedings show evidence reuse, because a plaintiff can reuse another regulator’s exhibit without showing one decision-maker read both. A shared incentive shows that each channel’s approver faced the same metric, not that the approvers were the same body.

Takeaway: Three provisions become one exposure only if one approver sits behind them, and the paper claims shared architecture only on evidence that names that approver.

II Meta’s Own Measurements Dated Its Awareness in Each Channel

Scam advertising produced the most precise file. The figures below are reported allegations, not adjudicated findings. Internal documents reported by Reuters on November 6, 2025 estimated 15 billion higher-risk scam ads a day and placed scam and banned-goods advertising at 10.1% of 2024 revenue, roughly $16 billion.

The same documents projected regulatory fines of up to $1 billion against roughly $3.5 billion earned every six months from high-risk scam ads. Rhode Island’s complaint in Providence Superior Court pleads the daily estimate and the revenue share under the state’s Deceptive Trade Practices Act.

Privacy produced the longest file. The Federal Trade Commission’s 2012 order and 2019 settlement each delivered findings in writing. On October 21, 2026, Judge Francis Mathew of New Mexico’s First Judicial District entered a $25 billion judgment after a jury found more than 43 million violations of state consumer protection law. New Mexico sued alone on the Cambridge Analytica conduct, and Meta has said it will appeal.

Child safety produced the file a jury read. Internal teen-safety research documented harms the product design sustained. The New Mexico jury found Meta liable for $375 million, and a judge was left to decide public-nuisance remedies in a later phase. On August 26, 2026 a coalition of 47 states and four other jurisdictions announced a resolution worth up to $17.1 billion of youth-safety and Cambridge Analytica privacy claims, with an independent auditor to assess implementation and efficacy.

Each channel holds a dated file, so awareness sits at Level 2a in all three. Receipt by a corrective authority reaches Level 2b in the scam-ad and privacy channels and remains open in child safety. MindCast’s prior twelve-entry forecast set on the teen-harm trial stands at one validated and none missed. The mid-trial resolution mooted four entries, two matched the outcome without a resolvable event and five remain open, as MCAI Lex Vision: Meta $16.68 Billion Settlement Validates MindCast Simulation Predictions — The Platform Design Liability Scorecard on the Mid-Trial Exit (2026) reports. A negotiated outcome removed the event the flagship entry targeted.

Takeaway: Awareness is established in each channel, so the contest turns on authority rather than awareness.

III Salience and Payoffs Explain the Shared Deferral

appears on every budget line a product owner tracks, while scam losses land on users and privacy harms land on no quarter’s results. Harm stays diffuse until an enforcer surfaces it, and a diffuse harm loses to a salient metric in each budget cycle.

Game theory supplies the payoff structure. In the scam-ad channel, Meta’s own documents projected fines of up to $1 billion against roughly $3.5 billion earned every six months. Comparable contribution figures are not public for privacy or child safety, so the payoff comparison rests on the scam-ad estimates alone.

The relevant comparison is not gross revenue against penalty totals. The comparison is the contribution Meta retained by keeping the practice, net of the correction’s cost, against the expected sanction discounted by the probability that an enforcer proves knowledge from the file. The deferral holds until a sanction reaches that contribution line, an authority outside the benefiting function takes the decision, or another route out of deferral opens.

The ratio is stark even before any discount. The documents put high-risk scam-ad revenue near $7 billion a year against fines of up to $1 billion, a gap of roughly seven to one. Discounting the fine by the odds that an enforcer proves knowledge widens the gap further.

Takeaway: In the one channel with public numbers, the expected sanction sat below the retained contribution.

IV The First Order Becomes the Next Measurement

The 2019 Federal Trade Commission settlement did more than fine Meta $5 billion. It required an independent assessor and routed privacy findings to a board committee. The board became the only body the public file shows receiving findings through a formal instrument.

The routing matters for the architecture question in Section V. Scam-ad findings reached the owners of revenue guardrails, and child-safety research reached recipients the public file does not name. If a shared approver exists, the board is the one body with documented receipt in at least one channel and authority over all three.

The Federal Trade Commission’s 2023 modification filing relied on the assessor’s findings, so the privacy reports already serve as evidence of what the committee received. Discovery showing the same committee or the full board receiving scam-ad or teen-safety findings would carry the paper’s claim from Level 2b to Level 3. Discovery showing those findings stopped below the board would confine each channel to its own approver.

Board-level receipt also changes the directors’ exposure. Bingle found no oversight liability where findings never reached the board, and an order that routes findings to the board removes that defense for the channel it covers. Meta’s directors hold documented receipt for privacy and an open question for the other two channels.

Takeaway: The privacy order made the board the body to watch, because it is the one recipient that could connect all three channels.

V The Authority Map Separates Shared Architecture From Shared Incentive

The authority map sets each channel’s enforcement event beside its internal finding and the recipient the public file names. The last column states the evidence that would raise the proof level. Unknown recipients stay unknown.

Three explanations compete for what the table shows. One decision rule may have governed all three channels through one approving body. Separate decision-makers may have responded to one shared incentive without a common approver. Independent technical and organizational failures may have produced similar outcomes at scale.

Awareness is established at Level 2a in each channel, and receipt at Level 2b in two. Shared incentive is an inference the mechanism supports, not a proof level. Shared architecture reaches Level 3 only if the Rhode Island discovery file or the New Mexico appellate file names one body that received findings from two or more channels and declined correction in two or more.

Two gaps separate what exists from what the public file can show. Confidential discovery or an early resolution can prevent observation of a shared approver without disproving one, and an appeal may add no new discovery. Construction Industry Laborers Pension Fund v. Bingle (Del. Ch. 2022) caps any enterprise-level claim that stops short of a named recipient.

A disclosed relocation of integrity approval outside the revenue line would establish a change in corrective authority. Level 3 would still require evidence that one body received findings and declined correction in multiple channels.

Takeaway: Shared architecture requires a named shared approver, and a relocation of approval would show changed authority without proving shared control.

VI Monetized Measurement and the Generator-Detector Conflict

Meta adds two configurations neither companion paper models. The first is monetized measurement. Reuters reported that advertisers Meta’s systems flagged as likely scammers paid higher rates through penalty bids rather than losing access. Whether the Rhode Island complaint pleads that practice remains open on the public file.

If substantiated, the practice turns the fraud signal into an income line. A declined measurement dates knowledge. A monetized measurement dates knowledge and the decision to profit from it.

The second is the generator-detector conflict. Rhode Island alleges that Meta’s systems helped create or optimize scam campaigns, and Meta’s AI ad-creation tools produce content its own detection must catch. Section VIII of the companion paper names four agentic failure shapes: alert suppression, the verification bottleneck, documentation outpacing performance and correlated blind spots.

The present configuration would add a fifth: an institution’s AI generates the load its review capacity must absorb. The fifth shape requires one fact the public file has not supplied, namely that the generation tools and the integrity function report to the same budget owner. The common reporting line remains a hypothesis, not established architecture.

Takeaway: Monetized measurement and generator-detector conflict each add evidence of incentive to evidence of knowledge, and each names an approval a plaintiff can date once discovery reaches it.

VII Two Comparators Test Whether the Pattern Is Meta’s Alone

UnitedHealth supplies a non-platform test. The February 2024 Change Healthcare cyberattack began on a server that lacked multifactor authentication, and the parent’s chief executive told Congress the company was still determining why. A November 2023 lawsuit alleged that more than 90% of coverage denials from the nH Predict algorithm were overturned on appeal. The two matters arose in separate units under separate law, and the parent held the budget to unify their controls.

Block supplies the fintech test. State regulators ordered Block to pay $80 million for anti-money-laundering failures on January 15, 2025, and the Consumer Financial Protection Bureau ordered $175 million the next day over fraud handling at Cash App. Both findings named one product, so correction authority over both sat inside one business.

A pattern shared by a health insurer and a payments firm points to incentive structure rather than business model. Neither comparator establishes a shared approver, so the comparison supports the behavioral mechanism and leaves the architectural claim to Meta’s own discovery file.

Takeaway: Shared incentive appears outside platforms, which supports the mechanism without proving shared control at Meta.

VIII States Will File in Parallel Rather Than Jointly

The solo route now shows a higher ceiling than the coalition route. New Mexico sued alone on Cambridge Analytica and holds a $25 billion judgment pending appeal. The 51-jurisdiction resolution of the same conduct and the youth-safety claims reached up to $17.1 billion, and Rhode Island’s share ran between $49.4 million and $64.5 million.

A headline judgment is not an expected payoff. The solo route bears appellate reduction risk, collection delay, litigation expense and a narrower remedy set. The coalition route delivered certain money, product changes and an independent efficacy auditor, so the comparison turns on the appellate outcome.

How State Attorney General Coalitions Form, Grow, and Hold (2026) supplies the formation conditions: lead-state fixed costs, portable evidence, compatible remedies and a coordination channel. The June 11, 2025 letter from 42 attorneys general on investment scam ads supplies the channel. Portable evidence now exists in Meta’s reported internal estimates, and Meta’s statement that Rhode Island retained an out-of-state firm shows how outside counsel lowers a state’s fixed cost.

The 14 separate TikTok suits of October 8, 2024 show the topology parallel filing produces. States coordinated on evidence and theory and filed in their own courts under their own statutes. Breadth appeared as a count of filings rather than a caption.

Parallel filing and joinder give states alternative responses under either appellate outcome in New Mexico. Meta’s litigation position depends more heavily on that single result. The 2025 letter therefore works as a coordination channel without producing a joint caption.

Takeaway: Enforcement breadth against Meta will appear as a filing count, because states can adjust their response to the appellate outcome while Meta’s position depends on it.

IX The Public File Favors Shared Incentive but Cannot Yet Show One Approver

Three explanations fit the evidence in the sections above, and each predicts a different discovery file. Each appears below with the evidence that would disprove it, before any forecast prints.

Hypothesis A: one approver: One body received findings in two or more channels and declined correction in two or more. Falsifier: public filings name separate approvers for each channel.

Hypothesis B: shared incentive, separate approvers: Each channel’s approver responded to the same revenue metric without common control. Falsifier: discovery names one approver across channels, or a channel’s approver funded correction against the revenue metric.

Hypothesis C: independent failures: Technical and organizational failures produced similar outcomes without a common incentive. Falsifier: internal documents show revenue guardrails bounding correction in more than one channel.

The public file supports Hypothesis B over C and cannot yet discriminate A from B. Hypothesis A therefore stays open without a band, and the Simulation Predictions below test propagation, disclosure and the evidence a court will accept.

X MindCast Simulation Predictions

The forecast set carries six Simulation Predictions: two primary (P) and four secondary (S). Two secondary entries are conditional and print as P(A given B). Verification sources and partial-outcome rules for each entry appear in Appendix B.

P-1 · A second U.S. state attorney general sues Meta over scam advertising by December 31, 2027 (67-79%)
A county, territory or private filing does not count. Falsifier: no second state files by the deadline.

P-2 · Meta publishes a scam-ad prevalence rate with its method by December 31, 2027 (30-43%)
Removal counts alone do not count, and a regulator-compelled figure does. Falsifier: no published prevalence rate by the deadline.

S-1 · Given P-1, the new filer sues on its own complaint rather than joining Rhode Island’s (74-86% given P-1)
A joint complaint among new states counts as a hit. Falsifier: the new filer joins or amends Rhode Island’s complaint.

S-2 · Given a ruling by December 31, 2027, the Rhode Island court denies Meta’s motion to dismiss the representation claims in whole or in part (60-76% given a ruling)
Falsifier: dismissal of the representation claims.

S-3 · A court ruling against Meta in one channel relies on an order, verdict or assessor finding from another channel as evidence of knowledge by December 31, 2028 (31-43%)
A filing that only cites such a document does not count. Falsifier: no ruling so relies by the deadline.

S-4 · The Rhode Island action resolves in full by negotiated judgment before a summary-judgment or trial ruling on liability by December 31, 2028 (29-42%)
A negotiated judgment after a motion-to-dismiss denial counts. Falsifier: a liability ruling, a litigated dismissal of all claims or no resolution by the deadline. The band is wide because Rhode Island’s per-violation ceiling is not yet public.

The six entries share one mechanism: evidence moves across forums faster than Meta’s approval structure changes. P-1 and S-1 test that propagation. P-2 and S-3 test disclosure and judicial reliance, while S-2 and S-4 trace the Rhode Island case.

Takeaway: The forecasts follow the evidence across forums, and the shared-approver question stays open until a named approver appears.

XI Risk Mitigation Layer

Exposure appears in a unit each stakeholder controls, with severity and probability stated separately. Entries sit together in one section, which keeps each audience’s actions in one place at the cost of distance from the register.

Platform integrity leads · P-2 (30-43%): Severity: high. Exposure: removal counts without prevalence rates become the plaintiff’s exhibit in discovery.

• Action 1: publish a prevalence measure with its sampling method, owner the integrity lead with approval from the executive who holds enforcement guardrails, before the Rhode Island motion-to-dismiss ruling.

• Action 2: file the method with the board committee that oversees integrity, owner the integrity lead, same deadline.

• Residual: a published prevalence figure dates the gap it measures.

Platform general counsel · S-3 (31-43%): Severity: high. Exposure: a ruling that relies on another channel’s findings widens discovery scope in every channel.

• Action 1: map approvals by body and channel for every open finding, owner the general counsel, within one quarter.

• Action 2: document which approver received each assessor finding, owner the chief legal officer, before discovery closes in Rhode Island.

• Residual: the map itself becomes discoverable.

State attorney general offices · P-1 (67-79%), S-1 (74-86% given P-1), S-4 (29-42%): Severity: moderate. Exposure: a coalition caption forgoes a per-violation ceiling, while a solo filing forgoes certain recovery and shared remedies.

• Action 1: plead the state statute individually and sign a common-interest agreement for evidence, owner the consumer protection chief, before the New Mexico appellate ruling.

• Action 2: evaluate the contingency arrangement against expected collectible recovery under both appellate outcomes, owner the attorney general, same deadline.

• Residual: an appellate reduction in New Mexico lowers the solo ceiling for every later filer.

Internal audit and insiders · S-3 (31-43%): Severity: high. Exposure: a finding that never reached an empowered approver proves awareness and not authority.

• Action 1: document each finding’s recipient and that recipient’s corrective power at delivery, owner the chief audit executive, each audit cycle.

• Action 2: escalate any finding whose recipient lacks funding authority to the committee that holds it, owner the chief audit executive, within the cycle.

• Residual: the escalation file dates knowledge at the top of the firm.

Institutional investors · S-2 (60-76% given a ruling), S-4 (29-42%): Severity: moderate. Exposure: three matters modeled as independent are one correlated position.

• Action 1: model Meta’s legal exposure as one correlated position across channels, owner the governance analyst, before the next annual report.

• Action 2: submit a disclosure question on approval authority for integrity enforcement, owner the stewardship lead, before the annual meeting.

• Residual: a denied motion to dismiss opens discovery whose contents no model captures in advance.

The actions share one direction. Each moves approval or documentation to a body that can act on the finding, and each leaves residual exposure because earlier measurements stay on file.

Takeaway: Mitigation narrows future exposure without erasing what the file already shows.

Conclusion

Meta’s 2026 judgments and filings test whether legal diversification diversifies governance risk. The answer turns on authority rather than motive. Where one body approved deferral in several channels, each enforcer inherits the last one’s evidence and the three matters become one exposure.

The evidence establishes dated awareness in each channel and receipt by a corrective authority in two. Discovery in Rhode Island and the New Mexico appellate file will show whether one approver sat behind each deferral. A shared approver supports the architectural claim, while separate approvers narrow the finding to a shared incentive.

P-1 tests propagation: a second state attorney general sues Meta over scam advertising by December 31, 2027 (67-79%). The architectural claim stays at Level 2a across channels and Level 2b in two until a named approver appears in more than one.

What to Watch

Checkpoints fall on January 21, 2027 and on October 23 of 2027 and 2028. The dominant fork is the Rhode Island motion-to-dismiss ruling.

Each signal maps to an entry. The Rhode Island ruling resolves S-2 and moves S-4. A second state filing resolves P-1 and S-1. Meta’s next integrity report resolves P-2, and the first ruling that relies on another channel’s findings resolves S-3.

Working With MindCast

Published analysis can map Meta’s three channels from the public file. It cannot tell a specific institution which of its own open findings reached an approver with power to fund the fix, or whether one approver sits behind deferrals in several channels. The answers sit in the institution’s measurement files, approval chains and board minutes, as the series introduction sets out.

MindCast builds Cognitive Digital Twins (CDTs) of an institution’s measurement, authority and verification system. It ingests the testing history and reruns the enforcement environment against the institution’s actual options.

For the question this installment asks, the engagement is an authority audit for boards, internal audit leaders and general counsel. The audit traces each open finding to its recipient and notes what that recipient could fund. It flags any approver who appears in more than one channel, which is the fact that converts separate matters into one exposure.

Commissioned work delivers three outputs. The first is a map of open findings by stage of conversion, from measurement to verified closure. The second is a set of exposure scenarios with low, median and high cases, and the third is a set of monitoring thresholds that show which scenario has begun.

The thresholds key to this installment’s Simulation Predictions. A second state filing (P-1), the Rhode Island motion-to-dismiss ruling (S-2) and the first ruling that relies on another channel’s findings (S-3) each mark a scenario shift. State enforcement teams can commission the same map from outside the firm, built on the public file and discovery.

MindCast takes one bounded next step at a time, and no engagement proceeds without a named decision to resolve. Contact mcai@mindcast-ai.com or visit www.mindcast-ai-simulation.com to schedule the diagnostic session.

Takeaway: The installment maps the public file, and an authority audit maps the institution’s own approval chain before a plaintiff does.

Sources

The works below support the facts and frameworks cited above. MindCast works come first, then primary sources, then secondary reporting.

MindCast Works

• The Foresight-to-Duty Conversion: Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found (2026). Supplies the stages, the second interval and the four agentic failure shapes.

• How State Attorney General Coalitions Form, Grow, and Hold (2026). Supplies the coalition formation conditions.

• Meta on Trial for Teen Social Media Harm: Twenty-Nine States, One Judge, and the Future of Platform Design Liability (2026). The prior forecast set on the teen-harm trial.

• Meta $16.68 Billion Settlement Validates MindCast Simulation Predictions — The Platform Design Liability Scorecard on the Mid-Trial Exit (2026). Assesses that set against the August 26, 2026 resolution.

Primary Sources

• Federal Trade Commission, In re Facebook, Inc. case page: 2012 Decision and Order; 2019 stipulated order; 2023 Order to Show Cause; July 30, 2025 stay of the show-cause proceeding.

• Rhode Island Office of the Attorney General, coalition resolution announcement, August 26, 2026.

• National Association of Attorneys General, letter of 42 attorneys general on investment scam ads, June 11, 2025.

• Construction Industry Laborers Pension Fund v. Bingle, C.A. No. 2021-0940-SG (Del. Ch. Sept. 6, 2022).

• Consumer Financial Protection Bureau, order against Block over Cash App fraud handling, January 16, 2025.

Secondary Reporting

• State of New Mexico v. Meta Platforms, Inc., First Judicial District, judgment of October 21, 2026, as reported by ABC News.

• State of Rhode Island v. Meta Platforms, Inc., Providence Superior Court, complaint filed October 22, 2026, as reported by PYMNTS.

• Reuters, special report on Meta’s internal scam-advertising documents, November 6, 2025, as summarized by MediaNama.

• New Mexico child-safety verdict, March 2026, as reported by NPR.

• Change Healthcare cyberattack testimony, May 1, 2024, as reported by the Associated Press via NY1.

• nH Predict lawsuit against UnitedHealth, November 2023, as reported by Healthcare Dive.

• Fourteen state suits against TikTok, October 8, 2024, as reported by ABC7.

Foresight-to-Duty Conversion Series Introduction: Meta Foresight-to-Duty Conversion Across Privacy, Child Safety and Scam Ads · Meta and Bank of America Foresight-to-Duty Conversion: Revenue-Funded Versus Cost-Funded Deferral · Consent Order Remedies at Meta, Boeing, Starling and Wells Fargo: Verified Correction Versus Recurrence

Related Works: Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much · The Foresight-to-Duty Conversion — Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found

Foundational Works: The Duty to Foresee, AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care · AI Governance Equilibrium · Decision Modeling and Foresight Simulation

Executive Summary

Enforcement can resolve a violation without changing the decision architecture that produced it. Effective assessment requires a timely corrective response and an operative consequence. External verification alone does not establish durable correction.

Thesis: A remedy changes the decision rule when a finding reaches someone who must respond on a fixed timetable and a consequence attaches if the response fails. The Federal Reserve lifted Wells Fargo’s asset cap on June 3, 2025 after third-party review and its own assessment. It terminated the 2018 action on March 5, 2026. Meta’s 2019 order supplied independent testing and agency oversight tools, and the corrective response stalled.

Boeing breached its 2021 deferred prosecution agreement, Starling breached a 2021 restriction on high-risk accounts, and TikTok agreed in August 2026 to pay $400 million to resolve alleged violations of its 2019 order.

What distinguishes the MindCast approach: The Foresight-to-Duty Conversion: Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found (2026) showed that the clock on inadequate governance starts when knowledge reaches an actor with authority. The present paper shows that a consent order can deliver that knowledge in writing, so a later stall on the specified deficiencies runs against a documented receipt. Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much (2026) supplies the live checkpoint through its S-2 entry on whether supervisors report detection-effectiveness results by October 13, 2028 (55-68%).

MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through the MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation (MP CDT FS) system. Behavioral economics explains why attestation fails: the function that writes the report answers to the authority that deferred. Game theory explains why order compliance raises the stakes: every report filed under an order becomes the next measurement file if the gap persists.

What the paper does: The paper classifies remedies by the powers they confer and the link where they stall, then dates knowledge under an order. It examines four firms that stalled again under supervision and one monitored bank whose later resolution tests the classification. It contrasts Wells Fargo’s relocated verification, defines a persistence test after oversight ends and closes with four Simulation Predictions and a Risk Mitigation Layer.

Simulation Prediction highlights: The forecast set carries four MindCast Foresight Simulation Predictions: two primary and two secondary. The lead entry forecasts that an enforcement action against a firm other than Meta cites that firm’s order compliance or assessor reports as knowledge evidence by December 31, 2028 (50-79%). The band is wide because the number of firms under order that will face a new action is uncertain.

🏛️ Policymakers: Remedies that pair effectiveness testing with a consequence the verifier’s finding triggers bind growth to correction. The Wells Fargo asset cap did so.

💼 Executives: Risk and compliance leaders: every assessment filed under an order dates knowledge for its recipients. Revenue owners: a remedy that leaves approval with your line leaves the decision rule intact. Boards: the link of the response chain a remedy leaves to management is where the next stall will sit.

🤖 AI deployment leaders: A launch gate with an assessor who can withhold sign-off attaches the consequence directly. A gate that only reports leaves the consequence to a regulator’s later action.

⚖️ Counsel: Compliance reports are future exhibits. A post-order action can cite them as evidence of knowledge for the recipients the order names.

📊 Investors: Termination of an order is an exit milestone, not proof of correction. New findings in the same domain during the three years after termination show which remedies held.

Across audiences the implication is the same. A remedy’s value lies in the response it compels and the consequence it attaches, not in the reports it produces.

I Remedies Fix Conduct, and Some Also Relocate Verification

A remedy names conduct and requires its cessation. Most remedies add a reporting obligation, and the firm’s own compliance function writes the report. Verification stays inside the authority that deferred.

A smaller class of remedies moves verification outside the firm and gives the verifier powers. Five powers matter. Independence from the firm and scope to test effectiveness come first. Access to evidence, escalation rights to the regulator and an enforcement consequence complete the set.

External location without those powers is attestation at a distance.

Available powers and delivered response are different things. Meta’s 2019 order installed an independent assessor with testing scope, reporting to a board privacy committee, agency oversight tools and penalties for false certification. The assessor found program gaps and reported them, and the agency’s response took the form of a 2023 modification proposal that the agency itself stayed in July 2025.

Classification follows two axes: the powers a remedy confers and the link in the response chain where it stalled. Meta held the powers and stalled at response. Boeing and Starling relied on reporting and the firm’s own controls.

Takeaway: The five powers describe what a remedy makes possible, and the response chain shows whether anyone used them in time.

II Consent Orders Change Deferral Payoffs Through Whom They Name, So Firms Fight Over Scope

A consent order changes the payoff of deferral more through its reporting terms than through its penalty. Before an order, an enforcer must prove that someone with authority received a finding. After an order, the instrument names the recipients and fixes the filing dates, so receipt of the deficiencies it specifies no longer needs separate proof.

The deferral inequality from the companion paper shows the mechanism. Deferral holds while the salience-weighted expected sanction stays below the cost of correction, and the expected sanction includes the probability that an enforcer proves knowledge from the file. An order pushes that probability toward certainty for the deficiencies it names. Deferral that was rational before the order stops being rational for those deficiencies, while deficiencies outside its scope keep their old odds.

Game theory therefore predicts where firms will spend their resistance. A firm facing an order gains more from narrowing its scope than from trimming its penalty, because scope sets the probability term for every later year. Meta’s contest of the Federal Trade Commission’s authority to modify its order fits the prediction, since the dispute concerns what the order may require rather than what Meta must pay.

TikTok’s August 2026 resolution fits it more sharply. TikTok agreed to pay $100 million more once its 2019 order is vacated, on top of $300 million for the alleged violations. Ending the order’s obligations carried a cost of its own, which is what the payoff structure predicts a firm will accept.

Behavioral economics explains why reporting terms bite before any enforcer acts. A finding addressed to a named committee or officer stops being a diffuse institutional fact and becomes a dated item in a specific person’s file. Salience rises for the person who must decide whether to fund the fix. Delaware oversight claims turn on whether a finding reached the board, as Bingle shows, so an order that routes findings to the board also routes exposure there.

The effect has limits the public file must respect. An order establishes notice only of the deficiencies it specifies, subject to its admissions and findings. A later assessor finding on a new gap starts its own interval when it reaches the recipient the order names, and a deficiency outside the order returns the enforcer to proving receipt from scratch.

AI deployers inherit the same structure. A launch gate or an order that names who receives evaluation findings converts each evaluation report into dated notice to a named person. The deployer then faces post-order odds on every finding the gate requires it to report.

Takeaway: An order changes deferral payoffs through whom it names and what it requires, so the fight over scope predicts how firms resist.

III Compliance Reports Become the Next Measurement File

Meta’s independent assessor found program gaps under the 2019 order, and the Federal Trade Commission’s 2023 modification filing relied on them. The report written to show compliance became the exhibit that showed its absence. The agency stayed its show-cause proceeding on July 30, 2025, and Meta contests the agency’s authority in federal court to modify a judgment the court entered.

Bank of America’s order sets up the same game for its peers. The order requires an independent review of the program. A bank under such terms can fund review capacity or narrow its documentation, and narrowing below what the review requires breaches the order’s terms.

Firms under order respond to the reports’ evidentiary value. A firm can describe process, meaning policies adopted and trainings completed. It can instead describe effectiveness, meaning detection rates and gaps closed. Process reports date knowledge of less, so a firm that keeps its decision rule drifts toward them.

Regulators counter by writing effectiveness testing into the order itself. Bank of America’s order requires an independent review of the program, and Section VIII treats the basis of the next follow-up as the live test of that contest. The balance between process reporting and effectiveness testing decides how much each report can later prove.

The reports also accumulate. A firm that files under an order for three years builds a dated series, and a plaintiff reads the series as a timeline of what the named recipients knew and when. P-1 forecasts whether an enforcement action against a firm other than Meta uses such a series as knowledge evidence.

Takeaway: Each report a firm files under an order can date knowledge for the next enforcer, so firms drift toward process reporting and regulators push back with effectiveness testing.

IV Four Firms Stalled Again Under Supervision

The paper separates three kinds of post-remedy event. New misconduct is conduct that began after the remedy. A continuing deficiency is a gap the remedy named that stayed open. Newly discovered historical conduct predates the remedy and says little about whether it worked.

The table classifies each remedy across the five links of the response chain, then names the stalled link and the type of post-remedy event.

Meta operated under an order from 2012 and a second from 2019. Assessor findings under the 2019 order showed continuing deficiencies, and the 2023 modification filing rested on them. The failed link was the corrective response, which has taken more than three years without a consequence attaching.

Boeing entered a deferred prosecution agreement in January 2021 without a monitor. On May 14, 2024 the Department of Justice found a breach for failing to design, implement and enforce a compliance program to prevent and detect fraud. The breach is a continuing deficiency in the required program, and the January 2024 door-plug failure brought scrutiny to it without itself constituting the breach.

The Boeing sequence continued after the breach. A court rejected a plea agreement in December 2024, a non-prosecution agreement followed in May 2025 and the court dismissed the charge in November 2025. The Fifth Circuit denied the victims’ challenges in 2026.

Starling Bank accepted a restriction on September 17, 2021 barring new accounts for high-risk customers. The Financial Conduct Authority’s September 27, 2024 final notice found that Starling opened 54,359 accounts for 49,183 such customers in breach of the restriction and fined it £28,959,426. The breach is new misconduct under a remedy the bank’s own controls enforced, and the notice found no formal program monitoring compliance with the restriction.

TikTok operated under a 2019 consent order entered against its predecessor Musical.ly. The Department of Justice sued in August 2024 for violating the order, and on August 21, 2026 TikTok agreed to pay $300 million plus $100 million once the 2019 order is vacated. The resolution includes no injunctive relief, so the order’s specific obligations end on vacatur while ordinary statutory enforcement continues.

Takeaway: Each second stall sits at a named link, and the classification separates new misconduct from continuing deficiency.

V A Monitored Bank Tests the Classification

Standard Chartered shows why the classification matters. The bank’s April 9, 2019 resolution of sanctions violations with U.S. and U.K. authorities extended its deferred prosecution agreements to April 2021. The bank stated that the vast majority of the conduct predated 2012 and none occurred after 2014, and both monitorships had ended by March 2019.

Most of that conduct is newly discovered historical conduct, which says little about the monitored remedy. The portion that continued into 2014 overlapped the period of supervision. A remedy with an embedded verifier can therefore coexist with ongoing violations for a time, and durability still has to be tested after exit.

Takeaway: Monitored remedies also require testing after oversight ends, and the classification keeps historical discovery from counting as recurrence.

VI Wells Fargo Shows Relocated Verification With a Consequence

capped assets until the board improved governance and risk management and a third party confirmed the improvements. The Board reviewed the third-party assessments and ran its own assessment before lifting the cap on June 3, 2025. It terminated the action on March 5, 2026 while other regulators’ orders continued.

The consequence was large in the currency management cared about. The Federal Reserve held Wells Fargo at $1.95 trillion in assets for about seven years, while Bank of America and Citigroup each grew total assets by 44% and JPMorgan Chase by 76%. The cap cost growth rather than cash, and growth was the metric on every senior executive’s line.

Behavioral economics explains why the design worked where fines had not. A fine lands once and spreads across shareholders, while a growth cap stays salient every quarter it binds. The verifier’s finding controlled the one consequence management could not budget around.

The case also limits the claim. The cap lifted after third-party and Federal Reserve assessments verified remediation, but durability after termination stays untested until the window in Section VII closes. Internal correction before an enforcer arrives remains a hypothesis, because no comparator on the public file documents it with enough precision.

Takeaway: Wells Fargo paired verification with a consequence denominated in growth, and the durability of that correction still awaits its post-termination window.

VII Persistence After Oversight Ends

Three endpoints require separate assessment. Procedural compliance covers policies, attestations and reports. Verified operational correction covers evidence that the deficiency closed. Durable institutional correction covers correction that persists after supervision recedes.

The test asks whether the remedy changed budget authority, escalation obligations or control performance, beyond adding an assessor. A remedy that changed none of the three left the decision rule intact.

The window is three years after the enforcement action terminates. Wells Fargo’s window runs from March 5, 2026 to March 5, 2029. Other provisions of the 2018 action stayed in force after the asset cap lifted, so the cap date does not start the window. Ordinary supervision continues through the window, and the test reads for findings classified as new misconduct or continuing deficiency.

Absence of a new action does not establish durability. Detection intensity falls after termination, and a gap can persist unobserved. Durability reads from two signals together: no classified finding in a covered domain within the window, and a disclosed change in budget authority or escalation obligation that survived termination.

Takeaway: Durability is tested after the action terminates, on classified findings and surviving structural changes, not on silence.

VIII Bank of America Is the Live Test

The Bank of America paper’s S-2 entry serves as the checkpoint: supervisors report detection-effectiveness results on remediation by October 13, 2028 (55-68%). A hit shows verified operational correction. It does not show durable correction, which requires the window in Section VII.

An effectiveness-based follow-up places Bank of America on the Wells Fargo path. An attestation-based follow-up places it with the firms in Section IV and tells peer banks that documentation suffices. Termination of the order would supply an exit milestone and start the window, not end the test.

Agency publication practice limits what the public file can show. Office of the Comptroller of the Currency terminations appear in monthly listings without narrative, and FinCEN releases rarely describe the testing behind a termination. Verified correction may occur without any public statement saying so.

Takeaway: The basis of the next supervisory follow-up classifies the remedy, though the public file may not show it.

IX Remedy Design for Regulators and AI Deployers

The cases in this installment support three design rules, each tied to the link where a remedy failed. Boeing and Starling relied on the firm’s own controls, so the first rule requires effectiveness testing by someone outside the firm. Meta’s assessor reported gaps that persisted for years, so the second rule sets a deadline for the corrective response to each finding.

Wells Fargo’s cap bound growth to the verifier’s finding, so the third rule attaches a consequence the finding triggers without a separate proceeding. The Wells Fargo case also shows the consequence works best when it lands on the metric management tracks.

The rules carry different costs. Effectiveness testing raises the cost of each order for the regulator. Deadlines raise the cost of delay for the firm, and verifier-triggered consequences shift bargaining power for the remedy’s whole life, which is why Section II predicts firms will resist them hardest.

For AI deployers the third rule takes the form of a deployment-readiness gate. The Federal Trade Commission’s 2023 proposal sought an assessor’s sign-off before new product launches at Meta, the control The Duty to Foresee: AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care (2026) described for autonomous systems. A gate works when the verifier can withhold sign-off directly, as the Federal Reserve’s control of Wells Fargo’s growth did.

Takeaway: Each design rule answers the link where a case failed, and a launch gate carries the third rule into AI deployment.

X Response and Consequence Compete With Verifier Powers as Explanations of Durability

The cases above admit three readings of why some remedies last. Each appears below as a hypothesis with its falsifier, and the comparison that would choose among them follows.

Hypothesis A: response and consequence predict durability: Remedies with a response deadline and an operative consequence show lower same-domain recurrence after termination. Falsifier: such remedies recur at the same rate as reporting-only remedies.

Hypothesis B: verifier powers alone predict durability: External verification with independence, scope and access is enough. Falsifier: remedies with full verifier powers recur when the response stalls, as Meta’s sequence suggests.

Hypothesis C: detection intensity drives observed recurrence: Recurrence reflects how closely regulators look rather than how well remedies work. Falsifier: recurrence rates stay stable across periods of differing supervisory intensity.

The comparative recurrence test remains open without a band. Among remedies imposed between 2019 and 2024 on the firms named here, none combines every link of the response chain with a completed three-year post-termination window, so the comparison group is empty until more cases close.

XI MindCast Simulation Predictions

The forecast set carries four Simulation Predictions: two primary (P) and two secondary (S). One secondary entry is conditional and prints as P(A given B). Verification sources and partial-outcome rules for each entry appear in Appendix B.

P-1 · An enforcement action against a firm other than Meta cites that firm’s order compliance or assessor reports as knowledge evidence by December 31, 2028 (50-79%)
A citation of the firm’s pre-order internal testing does not count. Falsifier: no such citation by the deadline.

P-2 · The Meta order modification resolves with a product-launch gate requiring independent assessor sign-off by December 31, 2028 (14-23%)
Withdrawal of the modification, a ruling barring it or no resolution by the deadline counts as a miss. Falsifier: no launch gate by the deadline.

S-1 · Wells Fargo faces a new federal enforcement finding of new misconduct or continuing deficiency in a domain its 2018 action covered by March 5, 2029 (17-32%)
Findings on conduct corrected before termination do not count. Falsifier: no classified finding by the deadline.

S-2 · Given termination of the Bank of America order by December 31, 2029, the termination statement describes effectiveness testing (17-33% given termination)
No termination by the deadline leaves the entry unresolved. Falsifier: termination without a statement describing effectiveness testing.

The four entries test the compliance-report mechanism and the durability window. P-1 tests whether reports filed under an order become evidence elsewhere. P-2 tests the launch gate, and S-1 and S-2 test Wells Fargo’s window and Bank of America’s exit.

Takeaway: The forecasts test whether reports become evidence and whether remedies hold after exit.

XII Risk Mitigation Layer

Exposure appears in a unit each stakeholder controls, with severity and probability stated separately.

Regulators · P-2 (14-23%), S-2 (17-33% given termination): Severity: high. Exposure: a reporting-only remedy signals to the regulated population that documentation suffices.

• Action 1: write effectiveness testing and a verifier-triggered consequence into each new order, owner the enforcement division, at the next order in the domain.

• Action 2: set a response deadline for each assessor finding, owner the enforcement division, same deadline.

• Action 3: publish the testing basis when an order terminates, owner the enforcement division, at each termination.

• Residual: a verifier with powers raises the cost of each order and lowers the number the agency can carry.

Boards of firms under order · P-1 (50-79%): Severity: high. Exposure: a plaintiff reads the series of compliance reports as a timeline of deferral.

• Action 1: fund the correction each report describes before the next report issues, owner the board committee the order names, each reporting cycle.

• Action 2: change the approval authority the report identifies as well as the control, owner the committee chair, before termination.

• Residual: the earlier reports remain dated knowledge of the period before correction.

Monitors and assessors · P-2 (14-23%): Severity: moderate. Exposure: findings without a response deadline or consequence become exhibits in a later action rather than corrections in the current one.

• Action 1: negotiate response deadlines and consequence terms into the engagement, owner the monitor, at engagement.

• Action 2: report effectiveness results rather than process compliance in each cycle, owner the monitor, each cycle.

• Residual: a regulator may decline the terms.

Consumer groups and public-interest researchers · P-1 (50-79%): Severity: moderate. Exposure: a recurring harm without a named approver and a dated finding stays a complaint rather than an accountability claim.

• Action 1: request the assessor reports and compliance certifications filed under each relevant order and map each reported gap to the recipient the order names, owner the research lead, before the next filing cycle.

• Action 2: classify each remedy in the domain by the response-chain link it supplied and publish the classification, owner the research lead, within two quarters.

• Residual: sealed assessor reports leave the recipient column open.

Investors · S-1 (17-32%), S-2 (17-33% given termination): Severity: moderate. Exposure: termination read as proof of correction understates risk during the three-year window when detection intensity falls.

• Action 1: hold the governance risk flag at its pre-termination level until a classified finding or a clean window resolves it, owner the governance analyst, at termination.

• Action 2: request disclosure of surviving budget-authority and escalation changes, owner the stewardship lead, before the first post-termination annual meeting.

• Residual: a clean window under reduced detection intensity still leaves an unobserved gap possible.

The actions share one direction. Each attaches a deadline or a consequence to a finding that would otherwise stay a report.

Takeaway: A remedy holds when the finding it produces can trigger action without a second proceeding.

Conclusion

Supervisory interventions at Meta, Boeing, Starling and TikTok preceded renewed stalls at different links of the response chain. Wells Fargo shows the alternative: a remedy that paired verification with a consequence the verifier’s finding controlled. Standard Chartered shows that even a monitored remedy needs testing after exit.

Correction still needs a durability test. Wells Fargo entered its three-year window on March 5, 2026. A classified finding in a covered domain before March 5, 2029 would move it from correction to recurrence. Bank of America’s next supervisory follow-up places it on one path or the other.

P-1 tests whether compliance reports filed under an order become evidence of knowledge: an enforcement action against a firm other than Meta cites that firm’s order compliance or assessor reports as knowledge evidence by December 31, 2028 (50-79%). Remedy durability rests on the persistence test in S-1.

What to Watch

Checkpoints fall on January 21, 2027 and on October 23 of 2027 and 2028. A final checkpoint falls on December 31, 2029. The dominant fork is the basis of the first Bank of America supervisory follow-up.

Each signal maps to an entry. The first post-order action citing compliance reports at a firm other than Meta resolves P-1. Any ruling or lifted stay in the Meta modification moves P-2. A classified Wells Fargo finding resolves S-1, and a Bank of America termination activates S-2.

Working With MindCast

Published analysis can classify remedies by the link where they stalled. It cannot tell a firm under order whether its own remedy changed the decision rule, or tell a regulator whether a remedy it is drafting will hold after exit. The answers sit in the order’s terms, the assessor’s reports and the firm’s approval chain, as the series introduction sets out.

MindCast builds Cognitive Digital Twins (CDTs) of the firm under order and its verifier. It ingests the reporting history and reruns the enforcement environment against the firm’s actual options before oversight ends.

For the question this installment asks, the engagement is a durability review for firms under order, their boards and their monitors. The review tests whether the remedy changed budget authority, escalation obligations or control performance, and it locates the link where the next stall would sit. Regulators and assessors can commission the same test on a draft remedy before it issues.

AI deployers face the remedy question as a launch-gate design question. A review-capacity simulation tests whether a gate would hold under the firm’s own evaluation volume, and whether the assessor’s sign-off carries a consequence or only a report.

Commissioned work delivers three outputs. The first is a map of open findings by stage of conversion. The second is a set of exposure scenarios with low, median and high cases. The third is a set of monitoring thresholds keyed to this installment’s Simulation Predictions: a post-order action citing compliance reports (P-1), the Meta modification (P-2) and the Wells Fargo window (S-1).

MindCast takes one bounded next step at a time, and no engagement proceeds without a named decision to resolve. Contact mcai@mindcast-ai.com or visit www.mindcast-ai-simulation.com to schedule the diagnostic session.

Takeaway: The installment classifies remedies from the public file, and a durability review tests one remedy before oversight ends.

Sources

The works below support the facts and frameworks cited above. MindCast works come first, then primary sources, then secondary reporting.

MindCast Works

• The Foresight-to-Duty Conversion: Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found (2026). Supplies the interval that dates a stall from receipt by an authority and the finding that measurement can become evidence against the institution that measures.

• Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much(2026). Supplies the S-2 checkpoint and the peer game.

• The Duty to Foresee: AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care (2026). Supplies the deployment-readiness gate.

Primary Sources

• Board of Governors of the Federal Reserve System, release of June 3, 2025 lifting the Wells Fargo asset cap, and release of March 5, 2026 terminating the 2018 action.

• United States Department of Justice, United States v. The Boeing Company case page: deferred prosecution agreement (January 7, 2021); breach determination (May 14, 2024); non-prosecution agreement (May 29, 2025); dismissal (November 6, 2025).

• Financial Conduct Authority, Final Notice to Starling Bank Limited, September 27, 2024.

• Standard Chartered, press release of April 9, 2019 on resolution of legacy sanctions conduct.

• Financial Crimes Enforcement Network, release on the Bank of America consent order, October 13, 2026.

Secondary Reporting

• Federal Trade Commission, In re Facebook, Inc. case page: 2019 stipulated order; 2023 Order to Show Cause; July 30, 2025 stay. Federal court contest as reported by MediaPost, February 16, 2026.

• United States Department of Justice resolution with TikTok and ByteDance, August 21, 2026, as summarized by Hunton Andrews Kurth.

The present series grows out of three MindCast papers. The Duty to Foresee: AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care (2026) argued that the duty of care an institution already owes can require it to assess foreseeable risk before deploying autonomous systems. Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much (2026) and The Foresight-to-Duty Conversion: Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found (2026) then showed what happens once an institution looks and keeps the gap it found: its own measurement becomes evidence against it.

• Duty-to-Foresee establishes the obligation to assess foreseeable risks before acting.

• Foresight-to-Duty Conversion describes how an institution’s measured knowledge and subsequent response become legally significant under an existing duty.

The new series takes up the questions those papers left open, namely why the response stalls, who controls it and whether enforcement changes it.

Executive Summary

Enforcement can resolve a violation without changing the decision that produced it. An institution that measures its own gap and keeps it holds a dated file, and the file becomes evidence once an enforcer reads it. The open question is whether the penalty, the monitor or the consent order changes who decides the next time the file shows a gap.

Thesis. The duty to foresee identifies when an applicable duty of care requires an institution to assess risk before it acts. Foresight-to-Duty Conversion explains what happens after it looks: the measurement becomes evidence of awareness, and the response decides whether that evidence becomes proof of inadequate oversight. The present series carries the conversion one step further and asks why the response stalls, who controls it and whether enforcement changes it.

What distinguishes the MindCast approach. MindCast AI applies Behavioral Economics + Dynamic Predictive Game Theory through the MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation (MP CDT FS). Behavioral economics supplies the decision rule: the metric a budget tracks stays salient while the harm stays diffuse. Game theory supplies the payoff structure: deferral holds while the expected sanction looks smaller than the cost of the fix. Predictive behavior emerges from the combination.

What the series does. Three installments each test one part of the decision. Installment I asks who controls correction when one platform faces several regulators. Installment II asks why institutions defer, comparing a revenue-funded deferral with a cost-funded one. Installment III asks whether a correction survives once oversight ends.

🏛️ Policymakers. Ask in every order who authorized the deferral and who verifies the fix. An order that answers both questions changes the next decision rather than only the last one.

💼 Executives. Every internal test, assessment and compliance report dates what your organization knows. Fund the correction a file reveals, or document a risk-based decline that survives a reckless-disregard reading.

⚖️ Counsel. Reconcile public statements against open internal findings before a plaintiff does. The representation channel reaches conduct that no statute names directly.

📊 Investors. Read remediation disclosures for detection results and approval changes, not spending totals. A disclosed change in who approves a fix carries more information than the penalty line.

I. The Duty to Foresee Established the First Duty

The Duty to Foresee opened the line of work that the present series extends. The paper matters here because it named the obligation every later installment tests: an existing duty of care can require an institution to look before deploying an autonomous system. It separated the duty to look from the duty to act and warned that foresight creates discoverable knowledge.

The paper rested the duty on Judge Learned Hand’s formula from United States v. Carroll Towing Co. (1947). Precaution is due when its burden falls below the expected harm it prevents.

B < P × L

B is the burden of the precaution, P the probability of harm and L its magnitude. The Duty to Foresee restated the expected harm for agent deployments as an incident rate times a loss magnitude, λ × L, and showed that the agentic version reduces to the classical Hand test when the agent-specific terms fall away. Foresight simulation supplies those inputs before deployment, so the duty to look becomes the duty to supply the formula’s terms.

Takeaway. The duty to foresee fixed when an institution must look. It left open what the look does to the institution’s legal position once it holds the answer.

II. Foresight-to-Duty Conversion Supplies the Mechanism Between the Duties

The Bank of America paper supplied the anchor case. The paper matters because it documented a bank that looked, measured and kept the gap. Bank of America’s 2022 test showed an unconstrained threshold would surface 85.4% of sampled suspicious activity against 20.2% under the existing rule, and management kept the rule. The Financial Crimes Enforcement Network (FinCEN) placed that unremediated test inside the willful course of conduct behind a $1.2 billion penalty.

The Foresight-to-Duty Conversion generalized the case into a framework. The paper matters because it supplied the mechanism the present series builds on: three stages, eight tests and three ways the response stalls. Conversion runs from receipt of the measurement by an actor with authority, to a funded decision, to verified closure. The paper tested the framework against sixteen enforcement cases and two controls, from TD Bank and Danske Bank to SolarWinds, and extended the Hand formula into time.

The extension measures the incentive an institution declines once its own file supplies the inputs. Let h be the expected harm per period that the precaution would prevent, and b the burden of the precaution per period. The accrued shortfall runs from the measurement at t₀:

D(t) = ∫ from t₀ to t of [h − b]⁺ ds

The bracket keeps only the positive part, so D accrues only in periods when the institution’s own numbers say precaution is due. Conversion completes at t*, when the gap is verified closed and D stops accruing.

The framework then states the private condition under which an institution defers. Let s be the salience assigned to a cost not yet paid, q the probability that an examination occurs and proves knowledge from the file, and F(D) the sanction as a function of D.

s × q × F(D) < b < h

The right inequality is the Hand condition: precaution is due. The left inequality is the private condition: the discounted enforcement cost looks smaller than the budget line. The gap between them is the deferral, and each budget cycle renews it with a larger D.

A second interval dates the shortfall from the moment someone who could act received the measurement, at tₐ:

D_c(t) = ∫ from tₐ to t of [h − b]⁺ ds

D dates awareness. D_c dates the stall under someone with authority, and the governing legal standard decides what each supports.

Takeaway. The companion papers explained how a measured gap becomes evidence. They did not explain why the response stalls, who controls it or whether enforcement changes it.

III. Why the Series Belongs to AI Governance

AI agents make measurement low cost and keep review scarce. An agent that surfaces more findings than a human review queue can clear produces dated knowledge at machine speed, and each scale-up gate renews the deferral the way a budget cycle does. The Foresight-to-Duty paper placed that acceleration at the center of its agentic section.

AI Governance Equilibrium (2026) defined governance debt as a liability that builds while autonomous activity outruns oversight capacity. The paper matters here because agents widen that gap faster than review capacity grows.

The first enforcement action that treats an AI agent’s pre-deployment evaluation as evidence of institutional knowledge will set the precedent for deployers. Banking, privacy and child-safety enforcement show how that action is likely to read the file. Each installment therefore carries one AI anchor alongside its case material.

Installment I examines Meta’s alleged use of AI ad-creation tools that produce content its own detection must catch. Installment II examines how AI changes the cost of review capacity, the variable behind Bank of America’s deferral. Installment III examines remedy design for AI deployers, where an assessor’s sign-off before launch works as a deployment-readiness gate.

Takeaway. The series studies the enforcement history that will govern AI deployers before the first AI case arrives.

IV. Three Causal Objects and Three Levels of Proof

The series separates three parts of one decision. The economic rule explains why an institution defers a correction. The authority structure explains who can preserve or correct the deferral. The verification system explains whether a correction endures.

Corrective authority and verification authority are different actors. The person who can fund a fix may differ from the person who judges whether it worked, and each installment keeps the two apart.

Every claim in the series sits on one of three levels of proof. Level 1 is an observed deficiency. Level 2 is demonstrated institutional knowledge of that deficiency. Level 3 is an established common decision architecture behind it, and each level requires its own evidence.

Takeaway. A deficiency does not prove knowledge, and knowledge does not prove a shared decision architecture. Each installment states the level its evidence reaches.

V. Enforcement Restarts the Decision Cycle

Enforcement rarely ends the decision cycle that produced a violation. An internal measurement leads to a deferral the institution’s economics favor. The deferral draws an intervention, and the intervention requires new compliance measurement.

The new measurement forces a renewed decision. Compliance reports, monitor findings and assessor reviews filed under an order can establish when corrective authorities received specific information about an unresolved deficiency. The governing legal standard decides what that receipt proves. If the decision rule survives the order, the next enforcer reads the institution’s own compliance file.

Takeaway. The series asks whether an intervention changes the incentives and authority behind the next decision, not only whether it resolves the last one.

VI. The Series Map

The series rests on three foundations and proceeds in three installments.

Foundations

• The Duty to Foresee: AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care (2026). The duty to look before acting, and the agentic Hand formula.

• Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much(2026). The anchor case of a measured gap kept by deliberate deferral.

• The Foresight-to-Duty Conversion: Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found (2026). The three stages, the extended Hand formula and the case framework.

Installments (available Mon Oct 26)

• Installment I. Meta Foresight-to-Duty Conversion Across Privacy, Child Safety and Scam Ads. Who controls correction when one platform faces several regulators.

• Installment II. Meta and Bank of America Foresight-to-Duty Conversion: Revenue-Funded Versus Cost-Funded Deferral. Whether funding source or corrective authority decides how long a measured gap lasts.

• Installment III. Consent Order Remedies at Meta, Boeing, Starling and Wells Fargo: Verified Correction Versus Recurrence. Whether enforcement changes the decision rule or only the conduct it named.

Each installment carries its own MindCast Foresight Simulation Predictions and a Risk Mitigation Layer. The series introduction carries none.

Takeaway. Read the foundations for the mechanism and the installments for the tests.

Conclusion

The duty to foresee told institutions to look. Foresight-to-Duty Conversion showed what looking does to an institution’s legal position once it keeps the gap. The present series asks why the gap stays open, who could close it and whether enforcement changes who decides next time.

The series makes one falsifiable claim at its center: enforcers will read compliance reports filed under a prior order as evidence of knowledge when the underlying gap persists. Installment III carries the Simulation Prediction that resolves the claim (band pending). The claim fails if later actions against firms with accessible compliance reports and persistent documented gaps set those reports aside as evidence of knowledge.

Sources

MindCast Works

• The Duty to Foresee: AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care (2026). Supplies the duty to look, the duty to act and the agentic Hand formula the series extends.

• Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much(2026). Supplies the anchor case, capacity-calibrated detection and the representation channel.

• The Foresight-to-Duty Conversion: Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found (2026). Supplies the three stages, the extended Hand formula, the deferral condition and the case framework.

• AI Governance Equilibrium (2026). Supplies governance debt as a liability that builds while autonomous activity outruns oversight capacity.

Legal Authority

• United States v. Carroll Towing Co., 159 F.2d 169 (2d Cir. 1947). The source of the Hand formula.

Working With MindCast

Published analysis can state the framework and map the series. It cannot tell a specific institution which of its open findings has already converted, who can authorize each correction or whether its last remedy changed the next decision. The answers sit in the institution’s own measurement files, approval chains and verification reports.

MindCast builds Cognitive Digital Twins (CDTs) of an institution’s measurement, authority and verification system. It ingests the testing history and reruns the enforcement environment against the institution’s actual options.

Commissioned work delivers three outputs. The first is a map of open findings by stage of conversion. The second is a set of exposure scenarios with low, median and high cases. The third is a set of monitoring thresholds that show which scenario has begun.

Four engagements follow from the series. For boards and compliance functions, an authority audit identifies who received each open finding and who can fund its correction (Installment I). For risk and finance leadership, a deferral review compares the expected return to deferral against the cost of each fix (Installment II). For firms under order and their monitors, a durability review tests whether the remedy changed the decision rule before oversight ends (Installment III).

For AI deployment leaders, a review-capacity simulation tests whether the human review queue can absorb what pre-deployment evaluation surfaces. The test runs before the first agentic precedent lands. MindCast takes one bounded next step at a time, and no engagement proceeds without a named decision to resolve. Contact mcai@mindcast-ai.com or visit www.mindcast-ai-simulation.com to schedule the diagnostic session.

Takeaway. The series supplies the framework, and a commissioned run applies it to one institution’s own files.

Companion works: The Economics Nobel Innovation Continuum (2025) · How Mozart Predicts Game Mutation in Predictive Behavioral Economics + Dynamic Game Theory (2026) · How Nietzsche Predicts Valuation Mutation in Predictive Behavioral Economics + Dynamic Game Theory (2026) · Chicago’s Randal Picker Was Simulating Law and Economics Before AI (2026)

I. Executive Summary

Core insight. An announced policy moves behavior today through two channels: what affected actors plan for explicitly and how they value what lies beyond their plans. A household that plans two quarters ahead weighs a rate promise for two years out through rough experience rather than explicit calculation. A long-horizon investor builds the same promise into its plans directly. Forecasts improve when they model how far each actor plans and how it evaluates consequences beyond that point.

The prize. The Royal Swedish Academy of Sciences awarded the 2026 economics prize to Robert E. Hall of Stanford University and Michael Woodford of Columbia University on October 12, 2026. The citation recognized “microeconomic foundations of macroeconomic stabilisation.” Central banks now run stabilization policy on their models while fighting energy-driven inflation alongside risks to growth. Regulators and legislatures face the same transmission question as they propose rules and pause approvals.

Thesis. The prize-winning framework treats expectations of future policy as the main channel of stabilization. Woodford’s own later research shows where the channel narrows: decision makers plan explicitly over a finite distance and evaluate the rest through learned experience. Horizon length therefore belongs beside credibility, incentives and capacity among the variables a forecast should model for each actor. Forecasts that assume unlimited foresight overstate the reach of distant announcements.

What distinguishes the MindCast approach. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through MindCast AI Proprietary Cognitive Digital Twin Foresight Simulations (MP CDT FS). Behavioral Economics supplies the decision rules. Game Theory supplies the payoff structure. Predictive simulations emerge from the combination.

Cognitive Digital Twins (CDTs) model each named actor from its observable conduct. Each CDT carries its own planning horizon rather than one horizon for the whole economy. A simulation can then run short-horizon and long-horizon actors against one another through the same announcement.

Expectations Horizon Vision. MindCast proposes Expectations Horizon Vision, a candidate Vision Function that forecasts which decisions each actor makes before an announced policy binds, and when. Mozart Vision forecasts when strategic games change their rules. Expectations Horizon Vision would estimate when an announced change reaches the actors inside the game. The function joins MindCast’s operating set once live cases validate it.

What the paper does. Sections II and III trace the prize-winning framework and the finite-horizon research that narrows it. Section IV specifies the proposed function, and Section V compares it with three established rival explanations. Section VI links the function to Mozart, Nietzsche and Picker Vision. Sections VII and VIII treat three live contests as cases to investigate and name the observations that would favor the horizon explanation.

• 🏛️ Policymakers: Effective dates inside the affected actors’ planning horizons carry more weight than promises beyond them. Match effective dates to the planning horizons of the actors a rule targets.

• 💼 Executives: Counterparties weigh distant commitments through rough experience rather than explicit plans. Deliver key commitments inside the counterparty’s horizon, such as first-year rate guarantees alongside ten-year tax projections.

• ⚖️ Counsel: Proposed rules and threatened suits weigh most on actors whose horizons reach the effective date. Assess each counterparty’s horizon alongside its legal exposure before treating a pending rule as a working constraint.

• 📊 Investors: Calendar promises about distant policy transmit weakly into near-term behavior. Build scenarios on what actors plan explicitly and how they discount the rest.

Takeaway. The 2026 prize confirms that expectations drive stabilization, and Expectations Horizon Vision proposes to estimate how far each actor’s expectations reach.

II. The 2026 Prize Rewarded Stabilization Built on Household and Firm Expectations

The 2026 prize recognized models that derive economy-wide stabilization from household and firm decisions. Hall’s “Stochastic Implications of the Life Cycle-Permanent Income Hypothesis: Theory and Evidence” (1978) showed that consumer spending reflects expectations about future income. Spending moves when expectations move. Hall also chaired the committee that dates U.S. recessions for decades.

Woodford’s Interest and Prices: Foundations of a Theory of Monetary Policy (2003) built monetary policy analysis on the optimizing decisions of households and firms. Expectations of future policy shape what households spend and what firms charge today. Central bank communication therefore carries real weight in the framework. NPR reports that the laureates’ work became the foundation of the New Keynesian approach to macroeconomics.

Woodford stated the governing idea after the announcement. He told reporters that authorities should not treat stabilization as an exercise in pulling levers without regard to what people in the economy understand. Prize committee chair John Hassler said the laureates gave central banks and governments a stable theoretical foundation for the models they now use.

Takeaway. The prize honors a framework in which expectations are the transmission channel of policy.

III. Finite Planning Horizons Explain Why Distant Promises Move Behavior Less

The standard New Keynesian model assumes rational expectations. Every household and firm forecasts the full future path of policy consistently with the model. Under that assumption a promise about distant policy has large effects on output today, and the effects grow with distance. Economists call the result the forward guidance puzzle, because actual central bank promises move economies far less.

Woodford’s “Monetary Policy Analysis When Planning Horizons Are Finite” (2019) addresses the gap directly. He models decision makers who look ahead only a finite distance. Beyond that distance they evaluate outcomes with a value function learned from experience. He drew the architecture from artificial intelligence programs for chess and go.

The finite-horizon model removes the forward guidance puzzle’s explosive effects. Woodford also shows that some well-known model results depend on applying rational expectations where they should be unreliable. Horizon length thus becomes a state variable of the forecast rather than a background assumption. Two actors facing the same announcement can respond differently when one plans two quarters ahead and the other plans two decades ahead.

Distant consequences still matter in the finite-horizon model. The learned value function carries them into present choices in compressed form, shaped by past experience rather than explicit calculation. A horizon therefore marks where explicit planning ends rather than where influence stops.

Takeaway. Finite horizons help explain why announced futures transmit unevenly, and the unevenness can be estimated actor by actor.

IV. Expectations Horizon Vision Is a Proposed Function for Estimating Each Actor’s Planning Reach

Expectations Horizon Vision asks two linked questions of every actor in a contest: how far forward does it plan explicitly, and how does it value what lies beyond? Inputs come from observable conduct such as contract durations, capital commitment timelines and litigation posture. Past responses to announced and enacted policy supply the strongest evidence. Horizon estimates come only from decisions made before the announcement, so the responses the function forecasts never feed its own inputs.

The output is a forecast for each actor: a named decision and the window in which the actor makes it. Examples include withdrawing an interconnection request, filing an enforcement action or changing a spending plan. A percentage of the eventual response appears only where evidence supports a measurable baseline, such as comparable actors’ responses once the policy binds.

Expectations Horizon Vision works through three questions. How far ahead does each actor plan explicitly? How heavily does it weigh consequences beyond that point? Who moves first once the announcement lands?

Game theory supplies the payoff structure and equilibrium selection, beginning with which actor gains from moving first when others discount the future. Behavioral economics supplies the horizon itself, along with bounded rationality and the salience that pulls near events into view. Predictive behavior emerges from the combination, because equilibrium selection changes once actors hold different horizons. A long-horizon actor facing short-horizon opponents can commit early and hold the position until its effects enter their plans.

The function rests on two assumptions. Actors reveal their horizons through conduct, and horizons change slowly relative to the announcements they filter. Horizon effects work alongside credibility, incentives, attention and capacity rather than in place of them. The observable test is forecast improvement: adding horizon estimates should sharpen forecasts of named decisions beyond what those factors explain.

Woodford’s finite-horizon planner and the CDT share a structure: limited lookahead joined to a learned evaluation of what lies beyond. Modern computational foresight makes that structure runnable for named actors rather than a representative household. MindCast Foresight Simulation Predictions built on the function would name the decision and response window for each actor an announced rule, ruling or commitment addresses.

Takeaway. Expectations Horizon Vision proposes actor-level forecasts of which decisions an announcement moves before it binds.

V. Credibility, Attention and Liquidity Compete With Planning Horizons to Explain Weak Transmission

Three established explanations also account for weak transmission of distant promises. Their footprints overlap in practice. Separating them requires comparing actors who differ on one dimension and resemble each other on the rest.

Imperfect credibility. Actors may doubt that the authority will deliver what it promises. Credibility can rise before implementation, so its footprint can resemble a horizon effect. The horizon explanation gains support when comparable actors facing the same credibility conditions respond in order of their planning horizons.

Inattention. N. Gregory Mankiw and Ricardo Reis’s “Sticky Information versus Sticky Prices” (2002) and Christopher Sims’s “Implications of Rational Inattention” (2003) hold that actors process news slowly. Attention can also rise as deadlines approach, which mimics a horizon effect. The horizon explanation gains support when actors known to have absorbed the announcement still respond in order of their horizons.

Liquidity constraints. Alisdair McKay, Emi Nakamura and Jón Steinsson’s “The Power of Forward Guidance Revisited” (2016) shows that households unable to borrow cannot act on promised future conditions. Capacity changes over time and can produce delayed responses that resemble horizon effects. The horizon explanation gains support when actors with ample capacity still respond in order of their horizons.

Xavier Gabaix’s “A Behavioral New Keynesian Model” (2020) shares the horizon mechanism’s direction through cognitive discounting of distant events. Expectations Horizon Vision contributes an application of the idea: it assigns named actors their own horizons from earlier conduct and tests them in live contests.

Expectations Horizon Vision fails where horizon estimates do not improve forecasts beyond credibility, attention, capacity and incentives. Comparable long-horizon and short-horizon actors would then make the same decisions in the same windows. The function would add nothing to the rival explanations in that case.

Takeaway. Rival explanations overlap with the horizon mechanism, so support for it comes from forecast gains beyond those rivals.

VI. Expectations Horizon Vision Supplies the Timing Variable for Mozart, Nietzsche and Picker Vision

The Mozart, Nietzsche and Picker papers each forecast a different kind of change. Expectations Horizon Vision adds the question all three leave open: when a change reaches each actor.

How Mozart Predicts Game Mutation in Predictive Behavioral Economics + Dynamic Game Theory (2026) introduced Mozart Vision to forecast why, when and how strategic games change their own rules. The paper separates anticipated changes such as petitions and proposed rules from realized ones. Anticipated changes move behavior before anything binds, and Expectations Horizon Vision would estimate how much. Mozart Vision’s Mutation Equilibrium turns on net gain “over the relevant horizon,” and the new function supplies that horizon for each actor.

How Nietzsche Predicts Valuation Mutation in Predictive Behavioral Economics + Dynamic Game Theory (2026) introduced Nietzsche Vision, which forecasts why valuations change, who makes them operative and which valuations carry into the next game. Horizon length shapes how heavily a revalued outcome weighs in a decision. A revaluation that becomes operative beyond an actor’s horizon enters its present choices only through learned experience, so Expectations Horizon Vision would estimate when it starts to carry full weight.

Chicago’s Randal Picker Was Simulating Law and Economics Before AI (2026) introduced Picker Vision, which identifies when locally rational interaction locks a population into one equilibrium and which minimum perturbation can move it to another. Heterogeneous horizons decide which members of a population respond first when an announcement arrives. Long-horizon actors can supply the early seed, and short-horizon actors follow once the effect enters their view.

The Economics Nobel Innovation Continuum (2025) read the 2025 prize to Joel Mokyr, Philippe Aghion and Peter Howitt as an account of how innovation generates growth. The 2026 prize turns from the sources of growth to how policy steers the economy around them. Read together, the two prizes cover the supply of new possibilities and the transmission of policy toward them.

Takeaway. Mozart Vision forecasts which game comes next and Nietzsche Vision which valuations carry into it. Expectations Horizon Vision would estimate when either change reaches present decisions.

VII. Three Live Contests Offer Cases for Testing Planning Horizons

Three contests underway in October 2026 offer cases for investigation. Each mixes horizon effects with differences in legal force, credibility and capacity. Each therefore calls for comparison across actors rather than inference from timing alone.

Prediction markets. On October 9, 2026 the Commodity Futures Trading Commission (CFTC) issued two instruments, both published in the Federal Register on October 19. An interim final rule took effect on publication and excludes casino-style gambling products such as sportsbook wagers from the definition of a swap. A proposed rule would add event contracts tied to sports, politics and other events to that definition, with comments due 30 days after publication. The instruments differ in substance and legal force as well as timing, so state responses cannot be read as horizon effects alone.

The useful comparison runs across states. State gaming regulators face the same federal instruments but differ in enforcement posture and revealed planning horizon. Expectations Horizon Vision would test whether states with longer horizons adjust enforcement before the proposed rule becomes final.

AI data centers. On August 3, 2026 Texas paused approvals for data centers in the Electric Reliability Council of Texas (ERCOT) interconnection process pending a verification and audit. On September 21 the Governor directed the Texas Commission on Environmental Quality to suspend data center permits until the ERCOT and Texas Water Development Board audits finish. Neither audit has a fixed end date. Hyperscale developers and local officials face the same pause with different capital commitments and plausibly different horizons.

The Two-Ledger Data Center Bargain (2026) showed that guarantees against feared losses outperform promised gains in contested sitings. Expectations Horizon Vision adds a timing hypothesis to that finding. Near-term guarantees enter local officials’ plans directly, while long-dated tax revenue weighs less. Billing cycles and election terms suggest those horizons but do not establish them, so the function would estimate them from past local decisions.

Central banks. Promises about the policy path several years out enter the explicit plans of long-horizon actors such as pension funds. Households and small firms weigh the same promises through rough experience, and liquidity limits and attention shape their response as well. Finite horizons suggest that guidance tied to incoming data reaches near-term spending more reliably than guidance tied to distant calendar dates.

• 🏛️ Policymakers: An interim final rule binds now while a proposed rule signals direction. Instrument choice changes both legal force and the set of actors that respond this year.

• 💼 Executives: Data center commitments that pay out in the first billing year enter local officials’ plans directly. Tax projections that pay out after a decade weigh less.

• ⚖️ Counsel: State enforcement against event contracts runs on its own clock while the proposed rule remains pending.

• 📊 Investors: Distant calendar guidance enters the plans of long-horizon institutions more directly than household spending plans.

Takeaway. Each contest pairs actors with different horizons under the same announcement, and comparison across those actors supplies the test.

VIII. Dated Events Will Show Which Observations Favor the Horizon Explanation

Five dated events through March 2027 supply cases for an incremental test. The test asks whether horizon estimates improve forecasts of each named decision beyond what credibility, attention, capacity and incentives explain. Earlier or simultaneous responses alone cannot identify the cause, because legal obligation and uncertainty also shape timing.

• November 9, 2026. Kalshi’s response to New Jersey’s petition is due in Flaherty v. KalshiEX, LLC (No. 26-299). Texas legislative prefiling also opens.

• About November 18, 2026. Comments close on the CFTC proposed rule. State enforcement actions before and after the deadline test whether horizon estimates sharpen forecasts of which states act first.

• Final Federal Open Market Committee meeting of 2026. The post-meeting statement shows whether guidance stays tied to incoming data or reaches for calendar dates.

• January 12, 2027. The Texas Legislature convenes. Developer withdrawals during the open-ended audits test whether horizon estimates improve forecasts of which firms withdraw.

• March 12, 2027. The Texas bill filing deadline arrives. Bills on large-load cost allocation show whether legislators act on long-dated costs before the audits finish.

The dominant fork. Forecasts of these decisions either improve when horizon estimates join the rival factors or they do not. Improvement across cases would support the function’s place in MindCast’s operating set. No improvement would mean the rival explanations already carry the timing.

Takeaway. Forecast improvement beyond the rival explanations is the evidence the function needs.

IX. Conclusion

The Royal Swedish Academy honored Hall and Woodford on October 12, 2026 for showing how expectations steer stabilization. Woodford’s finite-horizon research shows where explicit planning ends and learned evaluation takes over. Expectations Horizon Vision proposes to estimate that boundary for named actors across courts, agencies, legislatures and markets. Forecasts improve when they model how far each actor plans and how it values what lies beyond.

Takeaway. Forecasting an announcement’s effect starts with how far each affected actor plans ahead.

Working With MindCast

MindCast AI applies horizon analysis within its existing CDT simulations while Expectations Horizon Vision undergoes evaluation. An engagement estimates each actor’s planning horizon from its earlier decisions and forecasts which named decisions a pending announcement moves before it binds. The engagement then tests the client’s own commitments against those horizons before the client makes them.

• 🏛️ Policymakers test which effective dates move the affected population before drafting closes.

• 💼 Executives test whether a commitment lands inside the counterparty’s horizon.

• ⚖️ Counsel test how each pending rule or ruling weighs on opposing parties before it binds.

• 📊 Investors identify which actors respond to an announcement and which wait for binding force.

Takeaway. MindCast tests a strategy against the horizons of the actors it depends on. Engagements begin at mindcast-ai.com or mcai@mindcast-ai.com.

Sources

MindCast AI Works

• The Economics Nobel Innovation Continuum (2025). Supplies the 2025 prize reading that the 2026 vision extends from growth to stabilization.

• How Mozart Predicts Game Mutation in Predictive Behavioral Economics + Dynamic Game Theory (2026). Supplies Mozart Vision and the anticipated-change distinction that Expectations Horizon Vision quantifies.

• How Nietzsche Predicts Valuation Mutation in Predictive Behavioral Economics + Dynamic Game Theory (2026). Supplies Nietzsche Vision’s why-who-which account of valuation change, whose operative timing the new function dates.

• Chicago’s Randal Picker Was Simulating Law and Economics Before AI (2026). Supplies Picker Vision’s account of population lock-in and transition, which heterogeneous horizons sequence.

• The Two-Ledger Data Center Bargain (2026). Supplies the loss-weighted siting model that the data center contest extends with timing.

Primary and External Sources

• Royal Swedish Academy of Sciences. “Press release: The Prize in Economic Sciences 2026” (2026). Prize citation and laureate affiliations.

• Robert E. Hall. “Stochastic Implications of the Life Cycle-Permanent Income Hypothesis: Theory and Evidence.” Journal of Political Economy 86 (6): 971–987 (1978). Expectation-driven consumption.

• Michael Woodford. Interest and Prices: Foundations of a Theory of Monetary Policy. Princeton University Press (2003). The prize-winning framework.

• Michael Woodford. “Monetary Policy Analysis When Planning Horizons Are Finite”. NBER Macroeconomics Annual 2018, vol. 33. University of Chicago Press (2019). Finite-horizon planning.

• Martin Eichenbaum and Jonathan A. Parker. “Editorial”. NBER Macroeconomics Annual 2018, vol. 33 (2019). Statement of the forward guidance puzzle.

• N. Gregory Mankiw and Ricardo Reis. “Sticky Information versus Sticky Prices: A Proposal to Replace the New Keynesian Phillips Curve.” Quarterly Journal of Economics 117 (4) (2002). Inattention rival.

• Christopher A. Sims. “Implications of Rational Inattention.” Journal of Monetary Economics 50 (3) (2003). Inattention rival.

• Alisdair McKay, Emi Nakamura and Jón Steinsson. “The Power of Forward Guidance Revisited.” American Economic Review 106 (10) (2016). Liquidity rival.

• Xavier Gabaix. “A Behavioral New Keynesian Model.” American Economic Review 110 (8) (2020). Cognitive discounting.

• NPR. “2 Americans win Nobel Prize in economics for work on boom-bust cycles” (2026). Woodford’s post-announcement remarks.

• Axios. “U.S. economists win Nobel for work on managing economic shocks” (2026). Hall’s background and policy context.

• Al Jazeera. “Nobel economics prize awarded to Robert Hall and Michael Woodford” (2026). Committee chair statement.

• Commodity Futures Trading Commission. “CFTC Issues Interim Final Rule Excluding Certain Activity from the Definition of Swap”, Release No. 9309-26 (2026); 91 FR 65969. Casino-style gambling exclusion.

• Commodity Futures Trading Commission. Notice of Proposed Rulemaking on event contracts and the definition of swap, Release No. 9310-26 (2026); 91 FR 66022. Proposed inclusion of event contracts.

• Supreme Court of the United States. Flaherty v. KalshiEX, LLC, Docket No. 26-299. Petition filed September 2, 2026; response due November 9, 2026.

• Foley & Lardner. “Governor Abbott pauses Texas data center interconnections and calls for verification and audit”(2026). August 3 interconnection pause and open-ended audit.

• Troutman Pepper Locke. “Governor Abbott directs TCEQ to halt all data center permits pending ERCOT and TWDB audits” (2026). September 21 permit suspension pending audits.

• Texas Policy Research. “Legislative Calendar: 90th Legislative Session”. Prefiling, opening day and bill filing deadline.

Thesis. Courts, regulators, utilities and governments decide investment assumptions before earnings and ratings recognize them. MindCast AI simulates those institutions so investors can act inside the delay.

Why now. Five contests reach decision points between now and 2028. The Supreme Court holds petitions that decide whether event contracts become a national asset class, and Compass opened a second antitrust front against a multiple listing service (MLS) on October 16. Texas halted data center permits on September 21 with no size floor, and Tencent’s reported $7 billion lease of chips in Oracle data centers routes AI capability around the export gates of both Washington and Beijing. On October 13 Bank of America admitted a willful Bank Secrecy Act violation after its own 2022 tests had measured the blind spot regulators later penalized.

Who it reaches. Event-contract venues, brokerage shareholders and infrastructure lenders each meet one contest first. Cloud operators and bank boards meet the other two.

Contents

• I. Executive Summary

• II. The Supreme Court Decides Whether Event Contracts Become a National Asset Class

• III. Compass Carries Its Listing-Access Fight from Washington to California

• IV. One Texas Permit Halt Revalues Mid-Scale Compute

• V. Washington and Beijing Compete Through Two Sovereign Gates While Rented Compute Routes Around Both

• VI. Bank of America Measured Its Own Blind Spot, and the Measurement Became the Evidence

• VII. The Operative Instrument Connects Institutional Decisions to Investment Exposure

• VIII. MindCast Simulation Predictions

• IX. Risk Mitigation for Investors

• X. What to Watch

• XI. Conclusion

• Working With MindCast · Sources

I. Executive Summary

Core insight. Institutional decisions change investment assumptions before earnings, credit terms and ratings reflect them. The delay between the decision and its financial recognition is where long-horizon capital wins or loses. A permit halt changes the quarter a data center reaches revenue, and a court’s choice of statute decides which states a trading venue can serve.

Method. MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through MindCast AI Proprietary Cognitive Digital Twin Foresight Simulations (MP CDT FS). Behavioral Economics supplies the decision rules. Game Theory supplies the payoff structure. Predictive simulations emerge from the combination.

Original contribution. One mechanism repeats across five contests, from a Supreme Court case to a bank’s monitoring threshold: the authority that draws attention rarely binds first.

How to read the evidence. Linked sentences state verified facts and name their source. Bold probability bands mark MindCast Simulation Predictions, each with a window and a falsifier. Table notes flag MindCast’s own readings, and modeled payoffs are labeled as assumed.

The framework in brief. Each deciding institution becomes a Cognitive Digital Twin (CDT) built from its own filings, rulings and conduct. The CDTs play against one another until no actor can improve its position. Each resulting Simulation Prediction names an actor and an outcome with a window, a probability band and a falsifier.

Five analytical questions frame the cases:

• The changing-game question, applied to the Supreme Court contest, asks which game will exist next and which actor stays coherent when it arrives.

• The tipping question, applied to listing-service rules, asks when local decisions tip into a system-wide rule.

• The authorization question, applied to the Texas permit halt, asks which institution decides whether announced infrastructure becomes operating capacity.

• The sovereign-channel question, applied to the contest between Washington and Beijing, asks which sovereign action changes an investor’s exposure and through which channel.

• The duty-of-care question, applied to Bank of America’s monitoring threshold, asks whether an institution looked before it acted and governed to what it saw.

The Framework Companion sets out the method and the five questions in full.

The case studies. Five case studies carry the Opening Cases Companion. The table compares them on four dimensions: event, deciding institutions, MindCast finding and investor consequence. Each row stands alone.

What the Opening Cases Companion covers. Sections II through VI work through the five opening cases, and Section VII states the instrument that connects each case to investment exposure and the lessons each adds to MindCast’s framework. Section VIII sets out the series’ forty Simulation Predictions, including fifteen from the October 19 rented-compute papers and twelve from MindCast’s governance papers. Sections IX and X route six of them to the investors who carry the exposure and list the dated signals that resolve them.

• 📊 Investors: Underwrite on the institutional decision that binds first, not the one that draws headlines; in AI infrastructure that decision usually belongs to a utility or a state.

• 💼 Executives: Treat a regulator’s next move as a counterparty decision with a payoff structure you can model.

• ⚖️ Counsel: Contradictions across a litigant’s own filings in different forums predict outcomes better than docket counts.

• 🏛️ Policymakers: Rules copied across jurisdictions spread through tipping dynamics, so the first adopters shape the national regime.

Takeaway: in each of the five contests an institution changes an investment assumption before the financial result arrives.

II. The Supreme Court Decides Whether Event Contracts Become a National Asset Class

Prediction markets proved the demand for event contracts and left the hardest legal question open. The Intercontinental Exchange (ICE) committed up to $2 billion to Polymarket in October 2025 (ICE announcement). Kalshi reached a $22 billion valuation in early 2026, and 89% of its 2025 fee revenue came from sports. Whether sports event contracts count as federally regulated swaps or state-regulated bets now sits before the Supreme Court.

Both a Swap and a Bet: Simulating the Looming Supreme Court Battle Over Prediction Markets (August 21, 2026) holds that a single trade is both a swap and a bet across three layers: the venue, the contract and the conduct. Federal control of one layer need not reach the other two. Three outcomes compete before the Supreme Court.

• Full federal preemption. Federal commodities law governs the exchange, the contract terms and trading conduct, so states lose authority over sports contracts listed on a federally regulated exchange. Falsifier: the Court leaves any conduct regulation to the states.

• Full state authority. State gaming law reaches the contracts themselves, so federal registration does not protect the venue in restrictive states. Falsifier: the Court holds that federal law governs the exchange.

• Split layers. Federal authority governs the exchange while state authority governs conduct. Falsifier: the Court adopts either of the other two outcomes.

Both a Swap and a Bet names a split-layer ruling as the most likely outcome (55–60%). Under a split-layer ruling, sports revenue stays state by state rather than national.

From the MindCast executive presentation. Dashed borders mark Simulation Predictions; solid borders mark documented facts.

The changing-game question explains why the event-contract contest resists a clean win. The CFTC wants exchanges rather than sportsbooks, and states defend gaming revenue; 39 states joined a certiorari brief in October. Tribes enforce gaming law under a separate federal statute. On September 16, 2026 the Ninth Circuit held in Blue Lake Rancheria v. Kalshi that two California tribes are likely to succeed on federal Indian gaming law claims, adding a third sovereign to a two-sided fight.

MindCast’s prediction-market calls include a hit and a miss. Both a Swap and a Bet named the Ninth Circuit as the court likeliest to break with the Third Circuit, and the Ninth Circuit ruled seven days later. Both a Swap and a Bet also forecast a circuit count of 3–1 or 2–2 after the Sixth Circuit ruled; the count stood at 2–1, a miss.

📊 Investors: separate legal wins from operability, and weight event-contract exposure by state. Operators that keep both a prediction-market venue and a licensed sportsbook hold the hedge a split-layer ruling rewards.

Takeaway: the Supreme Court decides which states an event-contract venue can serve, and that map sets what a national event-contract business is worth.

III. Compass Carries Its Listing-Access Fight from Washington to California

Compass built a distribution strategy on private exclusive listings: homes marketed inside Compass before they reach the cooperative database rival brokers share. The Anywhere Real Estate merger assumed the private-listing premium would hold. Listing services defend the opposite rule, mandatory submission, under which a listing marketed to the public must enter the shared database within a set window.

Washington answered the listing-access question for one state, where the Northwest Multiple Listing Service (NWMLS) runs the shared database. The legislature passed Substitute Senate Bill (SSB) 6091 by 49–0 in the Senate and 92–1 in the House, barring marketing to a select group unless the home reaches the public and all brokers at the same time; MindCast’s founder testified on the bill. The Compass–NWMLS Settlement Changed the Private Governor After Washington Changed the Governing Law (September 1, 2026) found that mandatory sharing survived the August 31 agreement. The September 1 paper places Washington holding as cooperative absorption at 84%, a point estimate published without a closing date.

Compass opened a second front on October 16, 2026. Compass sued CRMLS in the Central District of California with Sherman Act counts that attack mandatory submission itself. Compass v. CRMLS Is the Suit That Tests It (October 17, 2026) observes that Compass endorsed mandatory submission in 2024 and preserved it in its Washington agreement. CRMLS had filed first in the Southern District of New York on October 5.

Two rules compete for the national listing market. A discriminating observation favors one reading without disproving the other, while a falsifier rules a reading out.

• Mandatory submission holds. Listing services keep requiring publicly marketed listings to enter the shared database, and peer services adopt the rule ahead of new statutes. Discriminating observation: CRMLS changes its rules to allow public marketing before submission.

• Private exclusives spread. Courts or agreements open a lawful path to market listings privately before submission, and other listing services follow. Discriminating observation: two or more peer listing services adopt mandatory-sharing reforms by August 31, 2027. Peer services can adopt different rules at once, so adoption favors the first reading without disproving the second.

The tipping question frames the national listing-access contest: mandatory submission is the incumbent rule across most of the country, and private exclusive listings are the rival rule. Washington reinforced the incumbent rule in one state, and California tests whether the rival rule can take hold in a second. Compass v. CRMLS places a further Compass antitrust complaint against a listing service other than CRMLS by the end of 2026 at 69–84%. Compass v. CRMLS also places a quiet rule change by at least one listing service that copies the California outcome at 40–54%.

From the MindCast executive presentation. Washington’s 84% figure is a point estimate as published.

The California litigation also reaches Compass’s equity. What the Private-Listing Economics Add to Magloire v. Compass, the Securities Act Case Already Pending Over the Anywhere Merger (October 17, 2026) covers the Securities Act suit over the Anywhere merger and places Compass’s third-quarter 10-Q naming both suits at 70–82%. MindCast’s Compass calls are mixed: the March 23, 2026 call that NWMLS would prevail or the case would resolve by agreement closed with the August 31 agreement, while two Compass timing calls missed.

⚖️ Counsel: Compass’s own 2024 endorsement and its Washington agreement both cut against the Sherman Act theory it now presses in California.

Takeaway: a publicly traded firm’s multi-state campaign against one industry rule becomes an equity question once merger synergies depend on the rule falling.

IV. One Texas Permit Halt Revalues Mid-Scale Compute

Owning land, storage and generation should let a clean energy company host AI compute first. Energy Vault broke ground on a host-owned AI campus in Snyder, Texas on July 27, 2026. The first phase places 10 Crusoe Spark modular data centers at the site with 8 megawatts of capacity, and a second phase could reach 25 megawatts. On September 21 the Texas governor directed the state environmental agency to halt data center permits pending grid and water audits, with no size floor.

MindCast ran thirteen CDTs on the Snyder case on October 9, 2026. The roster set four hyperscalers and two power providers alongside hosts, operators and state and county approvers. Lenders, firm clean power developers and curtailment providers completed it.

Two readings compete for which authority binds a host site first:

• The county binds first. Local boards react to the data center label and halt modules through zoning or moratoria. Discriminating observation: no county halts modules at an operating energy site before the end of 2027.

• The utility or state binds first. Load thresholds and state permits stop or delay a host site before any county acts. Discriminating observation: no host-sited modular project reports a delay from a utility or the state before the end of 2027.

Where the Clean Energy Host Advantage Holds and Where It Breaks in AI Data Centers (October 9, 2026) reports the Snyder result. The county, the authority that draws public attention, is the least likely of five authorities to bind a host site. A utility or the state binds first.

From the MindCast executive presentation. Route weights from the October 9, 2026 Snyder simulation describe host sites in general, not one project.

Behavioral economics explains why authorities split on host sites. County boards react to the salient “data center” label, while utilities apply load thresholds whatever the label says. Game theory explains the cost of the split: each authority writes terms inside its own instrument, and the host bears the cost of translating among them.

The Snyder simulation also found a squeeze on mid-scale AI load. Hyperscale campuses pass every gate by scale, and a single small module can sometimes connect below utility thresholds. Mid-scale load fails three gates at once: firm clean power suppliers set minimum contract sizes, federal reliability criteria reach only sites of 50 megawatts (MW) and above, and hearings judge the tier as a campus. MindCast places no hyperscaler hosting below 50 MW at a third-party energy site through 2027 at 70–82%, which shrinks the exit buyer pool for mid-scale capacity.

From the MindCast executive presentation. Solid cells mark documented rules; dashed chips mark Simulation Predictions.

The Snyder simulation’s paths set the capital decision. With assumed payoffs, staging funding to the executed utility service agreement beats committing on zoning once the chance that a utility or state binds the site tops 29%. The 55–65% route weight describes host sites in general, so a commissioned simulation sets each site’s own figure. Staging also keeps the decision reversible, which matters because Structural Intergenerational Behavioral Economics (January 23, 2026) finds that decision-makers stop optimizing once a commitment becomes irreversible and start preserving what they already hold.

📊 Investors: condition capital on the utility, not the county.

Takeaway: in AI infrastructure the authority that binds first is rarely the one in the news, and capital that waits for the binding approval keeps its bargaining power.

V. Washington and Beijing Compete Through Two Sovereign Gates While Rented Compute Routes Around Both

AI competition between Washington and Beijing now runs through permissions rather than chip performance. Revenue from a licensed chip sale requires two independent approvals: a United States export license and the buyer government’s acceptance. China’s H200 Import Block and the Reordering of National Innovation Control set out the two-gate mechanism in January 2026, and by May about ten Chinese buyers had cleared with zero chips delivered. While either gate stays shut, buyers qualify domestic suppliers, and requalification costs keep them there after the gate reopens.

From the MindCast executive presentation. Washington’s license and Beijing’s acceptance move independently, so revenue requires both.

Beijing’s refusal follows its own payoff logic. The two-gate paper found that large-scale H200 imports would anchor Chinese engineers to Nvidia’s software stack while domestic alternatives needed demand to mature. Game theory explains the payoff: refusing imports trades near-term capability for demand that domestic accelerators can grow into.

Rented compute now opens a third route around both sovereign gates. A lessee routes around Washington’s license because no rental license exists, and around Beijing’s acceptance because the chips never enter China. The Financial Times reported on September 30, 2026 that Tencent agreed to lease advanced AI chips across several Oracle data centers in Southeast Asia. The reported deal runs about $7 billion for roughly 100,000 chips over five years, and neither company has confirmed the terms.

Tencent, Oracle, and the Remote-Access Gap in US AI Chip Export Controls maps the gap between a chip license and a compute rental. A US rule or statute expressly reaching remote access takes effect by the end of 2027 at 30–49%, since a new Congress would start over once the 119th Congress ends on January 3, 2027. No US rule or statute forces existing offshore service to China-headquartered tenants to stop before October 31, 2027 at 88–95%. Behavioral economics explains why the gap persisted: regulators anchor on the salient object, and a chip crossing a border is salient while a training job is not.

Beijing wrote the same license-versus-rental distinction into its own rules. On October 16, 2026 China’s Ministry of Commerce and Ministry of Science and Technology proposed export controls on 40 technologies, 23 of them in AI. The proposal reaches model weights and computing clusters but exempts serving overseas users through apps, websites and application programming interfaces, according to the draft as reported. Both governments now leave remote service access outside the rules that govern technology transfer.

Three readings compete for Beijing’s stance on the Tencent lease, and each carries its own falsifier.

• Selective sovereign tolerance. Beijing deliberately permits offshore frontier training while protecting domestic accelerator demand at home. Falsifier: no official guidance ever distinguishes offshore training from domestic deployment.

• Corporate autonomy. Tencent pursues compute through a lawful commercial arrangement without a specific policy decision from Beijing. Falsifier: official guidance directing or restricting the lease, or similar leases appearing only among firms with strong state ties.

• Conditional tolerance. Beijing permits the arrangement while Washington imposes no general duty on operators to verify, disclose or restrict a tenant’s workloads, and changes stance once such a duty takes effect. Falsifier in the source paper: Beijing keeps silent after a US verification rule takes effect. Silence is consistent with informal guidance, so the observation discriminates among readings rather than disproving this one.

Existing end-use controls already reach part of what Washington can require of operators. A May 13, 2025 Bureau of Industry and Security policy statement says that supplying advanced chips to a data center provider that will use them to train AI models for a China-headquartered party may require a license under existing end-use rules, and that exporters may not self-blind to avoid the requirement. The proposed general remote-access regime would add a provider-side duty on top of those controls.

The MindCast simulation favors conditional tolerance from Beijing, exercised through informal guidance rather than formal law. A Chinese ministry publishing guidance that restricts foreign-operated compute within six months of a US verification rule sits at 18–38%.

The sovereign-channel question follows each sovereign action to the asset it reaches. A Canadian pension fund investing in an Ohio data center faces one sovereign instrument, while a Gulf sovereign fund investing in the same asset faces several moving together. The investor landscape simulation places new US chip export approvals for a Gulf AI vehicle such as MGX or G42 by June 30, 2027 at 64–76%. G42 and Core42 lose their export authorizations on April 6, 2027 unless BIS issues further notice.

How Export Control, Investment Screening, and Cable Geopolitics Reprice Data Center Capital places no study jurisdiction winning a binding US commitment on which compute it may receive by December 2028 at 80–90%. The same paper found that capital adapts in place far more often than it relocates.

Over the Tencent lease’s five-year term the question shifts from what Washington permits to whether Chinese labs still need Nvidia chips. A new or renewed China-parented offshore commitment of $1 billion or more surfacing during 2028 sits at 51–83%, the dependence outcome. Tencent, Alibaba or ByteDance stating by the end of 2029 that a flagship model was trained primarily on domestic accelerators sits at 32–53%, the substitution outcome. Both outcomes can hold at once.

American AI leadership depends on where frontier training runs and who can see it. Chinese labs renting from American operators pay American firms, train on Nvidia’s Compute Unified Device Architecture (CUDA) software stack and remain subject to any later BIS order that binds their operator. A rule requiring tenant identity and workload attestation keeps Chinese demand on American infrastructure while giving Washington a view of tenants and workloads it lacks today.

MindCast’s geopolitical calls include hits and a miss. The H200 delivery call holds on Nvidia’s reported figures, and the two-gate core holds through 2026. The call that Chinese hyperscalers would redirect workloads to domestic accelerators did not hold for training, because Tencent placed frontier training offshore.

📊 Investors: weight AI exposure by the sovereign instruments that reach each asset, and value offshore compute on the cash flows that survive a rule change. The Tencent–Oracle paper carries the rented-compute case in full, from the lease to operator markets and capital posture.

Takeaway: US–China AI competition runs through two sovereign gates, and rented compute routes around both until an operator rule closes the gap.

VI. Bank of America Measured Its Own Blind Spot, and the Measurement Became the Evidence

A compliance test protects an institution only if the institution acts on what the test finds. Bank of America’s monitoring system opened a case only when a customer’s alerts reached 20 risk points within 13 months, a threshold first set by stakeholder judgment. The bank’s own 2022 test found that an unconstrained threshold would surface 85.4% of suspicious activity in sampled below-threshold activity against 20.2% under the existing rule. Management kept the threshold, citing a practical concern about pushing too many event groups to cases.

The $1.2 billion penalty followed the bank’s own test file. On October 13, 2026 FinCEN assessed $1.2 billion with the OCC’s $750 million penalty credited toward it, and the bank admitted a willful violation of the Bank Secrecy Act. Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much (October 14, 2026) reports that the consent order quotes the bank’s own 2022 concern: “Yield may exhaust the investigation resources.” A 2024 lookback found nearly 8,900 unreported suspicious activity reports covering almost $4 billion and sized the deferred staffing at about 750 investigators.

From the MindCast executive presentation. Solid cells mark documented facts; dashed cells mark Simulation Predictions.

Four explanations compete for FinCEN’s finding, and each carries its own falsifier:

• Capacity calibration. The bank sized detection to the investigators it funded, and calibration files become routine enforcement evidence. Falsifier: later orders rest on findings that a program was absent without citing internal testing.

• Case selection. One customer tied to national-security exposure drove the penalty, so the finding stays specific to Bank of America. Falsifier: a peer order without national-security facts cites the same class of evidence.

• Technical defect. A system designed in 2010 failed regardless of staffing choices. Falsifier: remediation that adds investigators without replacing the system closes the detection gap.

• Null explanation. The order changes nothing outside Bank of America. Falsifier: any peer order or guidance that references threshold testing within the window.

The MindCast simulation favors capacity calibration. Behavioral economics explains the 2022 decision: case volume was the salient metric the budget tracked, while the cost of missed reports stayed invisible until an examiner surfaced it. An internal goal to raise the yield of suspicious activity reports per case turned efficiency into the target, and Goodhart’s law did the rest: a measure that becomes a target stops measuring what it was built to measure.

Game theory explains why the consent order reaches beyond one bank. The order creates no new duty, but it raises the cost of keeping unaddressed validation findings for every peer that holds them. Peers now choose between adding review capacity and narrowing what they document, and willful-blindness doctrine raises the cost of the second route.

The Foresight-to-Duty Conversion: Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found (October 15, 2026) generalizes the case across sixteen enforcement matters. USAA ran the inverse configuration, launching a replacement monitor that failed its parallel test and then overwhelmed its reviewers. TD Bank ran a budget rule its executives called the “flat cost paradigm” while about 92% of transaction volume went unmonitored. Willfulness in each case rested on knowledge rather than on the direction of the mismatch.

A public statement about controls opens a second channel the Bank Secrecy Act never provided. The bank lowered the 20-point threshold in December 2023 after engagement with its regulators, yet the consent order finds the monitoring deficiencies unresolved as recently as June 2025. The bank’s January 2025 anti-money laundering program letter described systems to identify and report suspicious activity while those deficiencies remained.

Under Omnicare, an opinion such as “reasonably designed” becomes actionable when the speaker omits facts that conflict with its basis, yet Singh v. Cigna treats generic compliance statements as puffery. A complaint therefore survives only where it ties a specific statement to a specific contrary test.

Bank of America’s decision structure reaches AI agents. An agent fleet that surfaces more findings than human reviewers can clear produces foreseen governance debt at machine speed, and every scale-up gate renews the decision Bank of America made in 2022. The Duty to Foresee: AI Deployment Readiness as Prospective Governance and the Arrival of Agentic Duty of Care (June 26, 2026) forecasts that pre-deployment simulation of agent behavior becomes an element of reasonable care by 2030 (70–80%), and the Bank of America order supplies a pre-agentic precedent rather than a confirming case.

From the MindCast executive presentation. Four failure shapes follow when agents surface findings faster than reviewers clear them.

MindCast’s earlier governance calls meet the consent order on mechanism, not on a resolved forecast. The Duty to Foresee warned in June that foresight manufactures discoverable knowledge, and the consent order cites testing reports from 2019, 2020 and 2022 as evidence of knowledge. The paper’s agentic forecast stays open, because the bank’s monitor runs on rules and points rather than machine learning.

📊 Investors: the direct penalty cost about 17 cents per share, while the bank’s own 2024 estimate of doubled investigation volume and 750 added staff points to a recurring remediation cost the penalty line does not show.

Takeaway: an institution that measures a gap and keeps it writes the exhibit an enforcer later reads, and remediation cost tells investors more than the penalty.

VII. The Operative Instrument Connects Institutional Decisions to Investment Exposure

Each of the five cases reaches capital through one operative instrument: the document or rule whose terms change cash flows when a named decision-maker acts. The instrument rarely matches the event that draws attention. Behavioral economics explains the gap, because observers anchor on the salient actor while the operative terms sit in a less visible document.

The table is MindCast’s reading of each case, drawn from the simulations in Sections II through VI.

Game theory explains why the gap between the salient event and the operative instrument persists. Every decision-maker sets terms within its own instrument, so the actor with the most visible platform often holds the least operative power over cash flows. Capital that tracks the operative instrument moves first, while capital that tracks the headline waits for a decision that changes less.

Four lessons from the cases carry into every later installment:

• Weight exposure by the operative instrument. Event-contract revenue follows state enforcement under a split-layer ruling, so state-level exposure measures the risk better than the headline ruling.

• Read a litigant’s filings across forums. A litigant’s positions in earlier forums, such as Compass’s 2024 support for mandatory submission, often say more about a new theory than the number of suits it files.

• Stage capital to the durable instrument. The Snyder simulation’s break-even, a 29% chance of utility or state control under assumed payoffs, shows when staging funds outperforms an early commitment.

• Read the measurement file before the penalty. Bank of America’s 2022 test, not the $1.2 billion headline, fixed what enforcers could prove and what remediation will cost.

MindCast’s four misses mark where the framework needs more structure. The circuit-count miss and two missed Compass timing calls concern court and litigation timing. The workload-shift miss assumed a hardware control would push Chinese demand toward domestic chips, while Tencent moved frontier training offshore instead.

📊 Investors: in each case the exposure that matters sits in a less visible instrument, so diligence should start with the decision-maker whose instrument sets the terms.

Takeaway: one operative instrument connects each institutional decision to investment exposure, and capital that tracks it keeps its leverage.

VIII. MindCast Simulation Predictions

The forty Simulation Predictions below come from the publications cited in each entry. P stands for Primary Simulation Prediction and S for Secondary Simulation Prediction, with identifiers taken from the source paper wherever it assigns one. All forty remain open as of October 20, 2026; resolved calls appear in Sections II through VI and in the Framework Companion.

Entries marked conditional apply only if their triggering instrument takes effect, and each resolves as not triggered rather than as a miss if the condition never occurs. Entries 13 through 27 come from the Tencent–Oracle and Cross-Border AI Compute papers of October 19. Entries 28 through 39 come from MindCast’s governance papers, entry 40 comes from the investor landscape simulation, and series installments add entries from 41 onward. The Bank of America paper’s S-3 resolves on a rubric rather than a band and stays in its source paper.

1. A split-layer ruling gives federal authority over the exchange and leaves conduct to the states (55–60%). Both a Swap and a Bet: Simulating the Looming Supreme Court Battle Over Prediction Markets. Window: resolves at the Supreme Court’s merits decision on the Kalshi petitions. Falsifier: the Supreme Court adopts full federal preemption or a full state win. Published August 21, 2026. Status: open.

2. The Supreme Court grants plenary review in Nos. 26-299, 26-338 or 26-344 by June 30, 2027 (64–72%).CFTC Seeks to Exclude Sportsbooks and Proposes to Claim Prediction Markets as Kalshi Petitions Reach the Supreme Court. Falsifier: no grant of review by that date. Published October 10, 2026. Status: open.

3. A tribal government raises federal Indian gaming law in a comment on either CFTC rule (82–90%). S-4, CFTC Seeks to Exclude Sportsbooks. Window: closes with the comment periods, 30 days after each rule’s Federal Register publication. Falsifier: no qualifying tribal comment by the closing dates. Published October 10, 2026. Status: open.

4. At least two peer listing services adopt mandatory-sharing reforms ahead of new state statutes by August 31, 2027 (60–75%). The Compass–NWMLS Settlement. Falsifier: fewer than two such adoptions by that date. Published September 1, 2026. Status: open.

5. Compass files an antitrust complaint against at least one listing service other than CRMLS by December 31, 2026 (69–84%). S-8, Compass v. CRMLS. Falsifier: no qualifying complaint against another listing service after October 17 and by December 31, 2026. Published October 17, 2026. Status: open.

6. CRMLS does not amend its rules to permit public marketing without submission between October 6, 2026 and June 30, 2027 (82–92%). S-6, Compass v. CRMLS. Falsifier: an amendment permitting that marketing. Published October 17, 2026. Status: open.

7. No county stops modules at an operating energy site through 2027 (58–70%). P-1, Where the Clean Energy Host Advantage Holds and Where It Breaks. Falsifier: a county order halts modules at an operating energy site. Published October 9, 2026. Status: open.

8. At least one host-sited modular project reports a utility or state delay by December 31, 2027 (62–74%). P-2, Clean Energy Host Advantage. Falsifier: no such report by that date. Published October 9, 2026. Status: open.

9. No major hyperscaler hosts compute below 50 MW at a third-party energy site through 2027 (70–82%). S-4, Clean Energy Host Advantage. Falsifier: a confirmed hyperscaler deployment below 50 MW at such a site. Published October 9, 2026. Status: open.

10. No study jurisdiction secures a binding US commitment on compute eligibility by December 2028 (80–90%).How Export Control, Investment Screening, and Cable Geopolitics Reprice Data Center Capital. Falsifier: a binding commitment announced for any study jurisdiction. Published August 12, 2026. Status: open.

11. A major rating agency takes a negative rating action on a data center financing that cites permitting, power delivery or local opposition by June 30, 2027 (30–42%). P-1, investor landscape simulation. Falsifier: no qualifying action from a qualifying agency by that date. Qualifying agencies: Moody’s · S&P Global · Fitch · KBRA · Morningstar DBRS. First published in the Framework Companion; simulation run October 18, 2026. Status: open.

12. A new US export authorization for advanced AI chips to a Gulf AI vehicle is announced by June 30, 2027 (64–76%). P-3, investor landscape simulation. Qualifying recipients: MGX, G42, Humain and the Qatar Investment Authority’s AI subsidiary. Falsifier: no new authorization announced in the window. First published in the Framework Companion; simulation run October 18, 2026. Status: open.

13. A US rule or statute expressly reaching remote access to controlled compute by foreign persons of concern takes effect by December 31, 2027 (30–49%). P-2 in Tencent, Oracle, and the Remote-Access Gap in US AI Chip Export Controls. Falsifier: no such instrument takes effect by that date. Published October 19, 2026. Status: open.

14. No US rule or statute forces existing offshore compute service to China-headquartered tenants to stop before October 31, 2027 (88–95%). P-1a, Tencent–Oracle paper. Falsifier: such an instrument takes effect before that date. Published October 19, 2026. Status: open.

15. A rating-agency report on a data center operator, project financing or asset-backed deal names tenant eligibility under US remote-access or export rules as a credit risk by December 31, 2028 (42–58%). G-S3, Cross-Border AI Compute. Falsifier: no qualifying report from Moody’s, S&P, Fitch or KBRA by that date. Published October 19, 2026. Status: open.

16. No public BIS letter or license condition stops the Tencent-Oracle service before October 31, 2027 (97–99%).P-1b, Tencent–Oracle paper. Falsifier: a public BIS letter or license condition that stops the service appears before that date. Published October 19, 2026. Status: open.

17. Conditional on entry 13: the first qualifying instrument imposes at least one provider-side tenant-identity or workload duty (74–88%). P-3, Tencent–Oracle paper. Falsifier: the first instrument relies only on a country list or entity list. If the condition never occurs, the entry resolves as not triggered rather than as a miss. Published October 19, 2026. Status: open.

18. Conditional on entry 13: the first qualifying instrument provides a transition, licensing or continuation path for contracts signed before its effective date (62–78%). S-1, Tencent–Oracle paper. Falsifier: the instrument applies to existing contracts on its effective date with no such path. If the condition never occurs, the entry resolves as not triggered rather than as a miss. Published October 19, 2026. Status: open.

19. A second China-parented offshore compute lease or commitment of $1 billion or more appears in major press reporting or a party disclosure by April 30, 2027 (58–82%). S-2, Tencent–Oracle paper. Falsifier: no such report or disclosure by that date. Published October 19, 2026. Status: open.

20. Conditional on a US tenant-verification rule taking effect by December 31, 2027: a Chinese ministry publishes guidance restricting foreign-operated compute outside China within six months (18–38%). S-3, Tencent–Oracle paper. Falsifier: no qualifying publication within six months of the rule. If the condition never occurs, the entry resolves as not triggered rather than as a miss. Published October 19, 2026. Status: open.

21. A new or renewed China-parented offshore compute commitment of $1 billion or more appears in major press reporting or a party disclosure during 2028 (51–83%). P-4, Tencent–Oracle paper. Falsifier: no such report or disclosure from January 1 to December 31, 2028. Published October 19, 2026. Status: open.

22. Tencent, Alibaba or ByteDance states by December 31, 2029 that a flagship model trained primarily on domestic accelerators (32–53%). P-5, Tencent–Oracle paper. Falsifier: no such statement by that date. Published October 19, 2026. Status: open.

23. Conditional on a qualifying US access instrument: a contract change attributed to tenant eligibility becomes public within 12 months of its effective date (58–71%). G-P1, Cross-Border AI Compute. Falsifier: no such public disclosure within 12 months. If the condition never occurs, the entry resolves as not triggered rather than as a miss. Published October 19, 2026. Status: open.

24. A specialist operator and a hyperscaler or cloud provider announce an arrangement assigning tenant-eligibility or remote-access compliance to one party by December 31, 2027 (25–35%). G-P2, Cross-Border AI Compute. Falsifier: no such announcement by that date. Published October 19, 2026. Status: open.

25. At least one US cloud provider offers audited or attested compute for tenants from restricted jurisdictions as a named service by December 31, 2028 (31–45%). G-P3, Cross-Border AI Compute. Falsifier: no named service by that date. Published October 19, 2026. Status: open.

26. Conditional on a qualifying US access instrument: the first instrument names tenant identity or workload logging as acceptable compliance evidence (65–79%). G-S1, Cross-Border AI Compute. Falsifier: the instrument names neither. If the condition never occurs, the entry resolves as not triggered rather than as a miss. Published October 19, 2026. Status: open.

27. At least one US modular or clean energy host publicly discloses tenant-eligibility terms in a host or operator agreement by December 31, 2027 (22–31%). G-S2, Cross-Border AI Compute. Falsifier: no such disclosure by that date. Published October 19, 2026. Status: open.

28. FinCEN or the OCC cites a second US bank’s own testing files as evidence of knowledge or reckless disregard by December 31, 2028 (62–74%). P-1, Bank of America $1.2 Billion Money Laundering Penalty and the Compliance System That Knew Too Much. The order targets a different US insured bank and expressly cites preexisting internal threshold tests, below-the-line sampling or model validation findings. Falsifier: no qualifying order text by the window close. Published October 14, 2026. Status: open.

29. FinCEN or the OCC issues a new major Bank Secrecy Act order requiring a below-threshold lookback or parameter-specific effectiveness testing by December 31, 2028 (66–78%). P-2, Bank of America paper. Qualifying orders carry a penalty of $100 million or more. Falsifier: no qualifying order. Published October 14, 2026. Status: open.

30. Shareholders file a derivative complaint naming Bank of America directors and citing the order’s internal warnings by October 13, 2027 (42–55%). P-3, Bank of America paper. Falsifier: no such filing by the window close. Published October 14, 2026. Status: open.

31. Shareholders file a securities class complaint targeting a specific Bank of America control statement by October 13, 2027 (25–38%). S-1, Bank of America paper. The complaint alleges that a specifically identified anti-money laundering control representation omitted contradictory testing or remediation information. Falsifier: no qualifying complaint, or one limited to generic risk-culture statements. Published October 14, 2026. Status: open.

32. FinCEN or the OCC publicly reports detection-effectiveness results on Bank of America’s remediation by October 13, 2028 (55–68%). S-2, Bank of America paper. Falsifier: the follow-up cites documentation only, or no public follow-up exists. Published October 14, 2026. Status: open.

33. Conditional on a FinCEN or OCC order by December 31, 2028 that cites a bank’s own testing or validation findings: the order also identifies the decision to defer remediation and the body or function that made it (55–68%). P-1, The Foresight-to-Duty Conversion: Measured Governance Debt Becomes Evidence When Institutions Keep the Gap They Found. If no qualifying order issues, the entry resolves as not triggered rather than as a miss. Falsifier: a qualifying order cites the testing findings without identifying the deferral decision or its locus. Published October 15, 2026. Status: open.

34. A US federal or state enforcement action against a deployer of an automated decision system or AI agent cites the deployer’s own pre-deployment evaluation or review-capacity records as evidence of knowledge by December 31, 2028 (58–70%). P-2, Foresight-to-Duty paper. Platform operators covered by MindCast’s platform-liability series are excluded. Falsifier: no qualifying action, or an action resting on outcomes alone. Published October 15, 2026. Status: open.

35. At least three of the ten largest US bank holding companies by total assets at June 30, 2026 disclose added financial-crimes investigative headcount or review capacity tied to transaction-monitoring remediation by December 31, 2027 (48–60%). P-3, Foresight-to-Duty paper. Falsifier: fewer than three qualifying disclosures, or disclosures attributing remediation to technology alone. Published October 15, 2026. Status: open.

36. A non-US financial regulator such as the Australian Transaction Reports and Analysis Centre or the UK Financial Conduct Authority issues an enforcement outcome citing an institution’s own capacity-calibration, alert-threshold or validation records as knowledge evidence by December 31, 2028 (62–74%). S-1, Foresight-to-Duty paper. Falsifier: no qualifying outcome, or outcomes citing program absence or results only. Published October 15, 2026. Status: open.

37. A securities or derivative complaint against a public company ties a specific public AI governance, safety or testing statement to a contradictory internal evaluation and survives a motion to dismiss on that statement by December 31, 2028 (40–52%). S-2, Foresight-to-Duty paper. Filing alone does not qualify. Falsifier: no complaint survives on a specific statement, or dismissals rest on puffery or generic risk-factor grounds only. Published October 15, 2026. Status: open.

38. A Securities and Exchange Commission enforcement action under the 2023 cybersecurity disclosure rule cites a registrant’s internal alert-suppression, triage-capacity or unreviewed-alert records as evidence bearing on a disclosure’s accuracy by December 31, 2028 (33–45%). S-3, Foresight-to-Duty paper. An action resting on the timing of an incident filing alone does not qualify. Falsifier: no qualifying action. Published October 15, 2026. Status: open.

39. At least one regulatory regime, professional standard or judicial decision treats failure to simulate or stress-test an autonomous agent system before deployment as evidence of unreasonable conduct or inadequate oversight by December 31, 2030 (70–80%). Forecast, The Duty to Foresee: AI Deployment Readiness as Prospective Governance and the Arrival of Agentic Duty of Care. Scope: enterprises deploying autonomous agents into operations affecting third parties, shareholders or critical systems. Falsifier: agent-harm disputes continue to resolve on capability, defect or misuse theories with no foresight-precaution element. Published June 26, 2026. Status: open.

40. Citadel Securities becomes a confirmed market maker in non-sports event contracts on a venue regulated by the Commodity Futures Trading Commission by June 30, 2027 (50–62%). P-2, investor landscape simulation. Falsifier: no public confirmation by the venue or the firm by that date. First published in the Framework Companion; simulation run October 18, 2026. Status: open.

Takeaway: each Simulation Prediction names who acts and what happens, the date or event that closes its window and what would prove it wrong.

IX. Risk Mitigation for Investors

The entries below route six Simulation Predictions from Section VIII to the audience that carries the exposure. Severity and probability appear as separate axes. Nothing in the section is investment, legal or fiduciary advice.

Entry 8, for infrastructure lenders: a utility or state delays a host-sited project (62–74%).

• Exposure: quarters of delay to first operation, carried on equipment deposits paid before service is confirmed. Severity rises when the delay crosses a refinancing date.

• Mitigating actions: credit committees make import confirmation and a permit inventory conditions of the first draw by the next term sheet. Deal teams tie later draws to the executed utility service agreement before the Texas grid audit lands, which the grid operator targets for December 10, 2026. Counsel writes approval-delay allocation into each host agreement before signing.

• Residual exposure: a state halt with no size floor can still reach a site with every utility approval in hand.

Entry 1, for event-contract venues and market makers: a split-layer ruling (55–60%).

• Exposure: revenue confined to permissive states, measured in the share of volume from states that sustain restrictions.

• Mitigating actions: strategy teams build state-level revenue forecasts before the Court acts on the petitions. Product teams prepare geolocation exclusion for Indian lands ahead of a final CFTC rule.

• Residual exposure: a ruling that reaches contract design as well as conduct would narrow the product set itself.

Entry 5, for holders of residential brokerage equity: a Compass complaint against a listing service beyond CRMLS by year-end (69–84%).

• Exposure: merger synergies that assume private-listing economics survive, measured in the share of projected revenue tied to private exclusives.

• Mitigating actions: analysts test synergy assumptions against mandatory submission state by state before the third-quarter 10-Q. Portfolio managers read the forum ruling in New York as the first resolution of the California dispute.

• Residual exposure: a negotiated marketing window in California would leave the rule standing while changing its cost.

Entry 18, for cloud operators and their lenders: the first remote-access rule provides a transition path (62–78%), conditional on a rule taking effect.

• Exposure: contracted revenue from China-parented tenants, measured in the share of lease cash flows signed before a rule’s effective date.

• Mitigating actions: general counsel adds transition, suspension and refund terms to every new offshore compute lease before October 31, 2027. Operator leadership documents tenant-identity and workload controls that can be offered as license conditions. Credit committees size exposure to China-parented tenancy against both a tenant test and a country list.

• Residual exposure: a rule without a transition path overrides contract terms, and intermediaries can mask the tenant a lease names.

Entry 28, for peer bank compliance chiefs: FinCEN or the OCC cites a second bank’s own testing files as evidence of knowledge (62–74%).

• Exposure: discovery reaches every validation report and below-threshold test since 2018. Bank of America’s own estimate of 750 added investigators is the benchmark examiners now carry, while each peer’s need depends on its own volumes.

• Mitigating actions: compliance functions inventory open validation findings within 30 days. Model risk teams rerun below-threshold tests on current thresholds within 90 days, and the compliance chief presents a funded closure plan to the board risk committee before the next examination cycle.

• Residual exposure: historical files remain discoverable, and closure plans document knowledge of the gap they close.

Entry 35, for investors in large US bank holding companies: three of the ten largest disclose added investigative capacity by the end of 2027 (48–60%).

• Exposure: recurring staffing commitments persist beyond the penalty year as an operating line, and banks that blur them with one-time remediation spending leave the deferred headcount unknown.

• Mitigating actions: analysts add deferred review capacity to the operating-cost line as a recurring cost, not a reserve, before the next earnings date. Models release that cost only on a disclosure that states a detection result rather than a spending figure.

• Residual exposure: a bank that funds quietly leaves no disclosure to read, so models understate closure for banks that remediate without saying so.

Takeaway: each exposure has an owner, a deadline and a residual that no action removes.

X. What to Watch

• October 19, 2026: the Texas environmental agency’s report on the permit halt fell due, and its text shows whether the halt reaches existing energy sites.

• November 3, 2026: state elections, which MindCast expects to shift the partisan ceiling of attorney general coalitions by one to three offices (78–88%).

• November 15, 2026: Beijing’s consultation on export controls for 40 technologies closes, and final text shows its stance on service access.

• December 10, 2026: the Electric Reliability Council of Texas targets this date for filing the grid audit that sets the exit from the permit halt.

• December 31, 2026: the window closes for a Compass complaint against a listing service beyond CRMLS.

• Before January 3, 2027: the defense authorization conference report shows whether remote-access text survives before the 119th Congress ends.

• April 6, 2027: G42 and Core42 export authorizations expire automatically unless BIS issues further notice.

• April 30, 2027: the window closes for a second China-parented offshore compute commitment of $1 billion or more (58–82%).

• June 30, 2027: the windows close for plenary review of the Kalshi petitions, a rating action citing authorization and new Gulf chip authorizations.

• Through October 13, 2027: Delaware Chancery and federal dockets show whether shareholders file a derivative complaint naming Bank of America directors (42–55%) or a securities complaint on a specific control statement (25–38%).

• October 31, 2027: the windows close for a forced stop of existing offshore compute service and for a public BIS stop order against the Tencent-Oracle service.

• December 31, 2027: the windows close for a US remote-access rule (30–49%), a specialist-hyperscaler compliance arrangement (25–35%) and a host disclosure of tenant-eligibility terms (22–31%).

• December 31, 2027: the window closes for three of the ten largest US bank holding companies to disclose added investigative capacity tied to monitoring remediation (48–60%).

• During 2028: new or renewed China-parented offshore commitments show whether dependence on rented compute persists.

• October 13, 2028: the window closes for a public FinCEN or OCC report of detection-effectiveness results on Bank of America’s remediation (55–68%).

• December 31, 2028: the windows close for a named audited-compute service from a US cloud provider (31–45%) and a rating-agency report naming tenant eligibility as a credit risk (42–58%).

• December 31, 2028: FinCEN and OCC order text shows whether a second bank’s own testing files become knowledge evidence (62–74%) and whether a new major order mandates a below-threshold lookback (66–78%). The same date closes the window for enforcement citing an AI deployer’s own evaluations (58–70%).

• Through 2029: a Tencent, Alibaba or ByteDance statement on a flagship model trained on domestic chips would show substitution.

• Through 2030: a regulator, standard or court treating skipped pre-deployment agent simulation as evidence of unreasonable conduct would confirm the duty to foresee (70–80%).